STRC: An Assessment of Buybacks, Market Structure & Channel Sensitivity

“The goal is to turn data into information, and information into insight.”

Summary

Stretch (STRC) closed at $97.75 on September 4, 2026, having recovered from an intraday low of $71.25 on June 26. Across 278 daily closes from inception to 9-4-2026, STRC recorded 35 days [12.6% of trading sessions] closed at or above par; with another 190 sessions [68.3%] closed between $95-$99.99. In total, in 225 of 278 sessions [80.9%] closed at $95 or higher.

Over the period of June 11–September 4, a 60-Session orthogonal attribution model explains [74.65%] of daily-return variance in STRC allocating [44.7%] of the explained portion to the other market Digital Credit preferreds as a block, [39.2%] to MSTR excluding Bitcoin, [15.7%] to Bitcoin, and less than [0.5%] combined to Trading Volume and Daily Short Volume Flow. The orthogonal design run on STRK | STRF | STRD and SATA finds the preferred block has the largest allocation in all five models with an attribution range of [44.7%] to [66.7%].

Between July 20 and September 4, Strategy repurchased 8,541,087 STRC shares [~$811.5M] or [17.43%] of share volume; over that period, reported short-sale volume was of roughly the same scale [16.96%]. Here we should be careful to note that activity share is not the same as a variance share: residualized trading volume | short-volume flow were not statistically distinguishable from zero in the 60-session model (HAC p = 0.37 | 0.73), and a session-allocated buyback proxy adds 0.106 percentage points of fit with (HAC p = 0.50) indicating that the result is not statistically significant.

All Strategy preferreds rebounded between June 26 and July 17, before the first disclosed repurchase, and there were no reported repurchases of STRK | STRF | STRD at all. Evidence supports the view that the recovery was part of a broader repricing already underway across the Digital Credit complex, and repurchase execution may not have been necessary. A separate, inferential reading of the speed | depth of the selloff in June (that leverage built up against the instrument’s scale and engineered par anchor) is proposed in the Market Read section. The open question is not whether Strategy can keep buying, but whether they should, and how much organic demand remains once they stop.

Introduction

While instruments like STRC, STRF, STRK, STRD and SATA are classified as perpetual preferred equity, digital credit, or high-yield assets that sit senior to their issuer’s common stock, those labels do not fully describe how they trade. At their core, they have functioned as a} economic derivatives of Bitcoin with a yield feature, b} claims linked to their Parent Equity, c} reflexive layers of the broader Digital Credit complex. Any weakness in Bitcoin, real or perceived, transmits through the “base” of the entire capital structure, just as parent equity weakness can spread through the same levels. Structural features such as dividend rates, cash reserves, rate resets, and share buyback programs can (in theory) mitigate these effects, but they do not erase the broader transmission channels.

One can argue that this dynamic is central to their design; with Bitcoin acting as the foundational layer, the parent equity as the amplified wrapper on that layer, and credit instruments serving as secondary rails built upon both. Though tailored for different durations, seniority tiers, and investor types, they share the same fundamental exposure, functioning both as shock absorbers for, and beneficiaries of, Bitcoin and Parent Equity volatility. It is probably helpful to think of Digital Credit instruments as Meta-Derivatives: they are BTC Derivatives, and they are Derivatives of their Parent Equity that is itself a BTC Derivative (a derivative of a derivative). This understanding is essential if we are to properly analyze their performance over time.

As the largest of these credit rails, Stretch STRC has been the main subject of a recent debate primarily focused on whether a corporate bid (and other corporate adaptations) can allow a meta-derivative to trade closer to a pure fixed income product or money market fund. The following analysis will assess STRC from a variety of angles with a focus on performance, trading dynamics, inherent sensitivities, share buybacks, the role of leverage, and a variance attribution model that shows factors associated with return variance across the Digital Credit complex.

STRC Performance | Price Range(s): Data Through September 4, 2026 | Repurchase Disclosures Through September 8, 2026

Stretch (STRC) is marketed as a preferred offering that “strips away” volatility, which naturally raises the question: relative to what? Considering Strategy’s capital structure and its framing of STRC alongside other Digital Credit instruments as “overcollateralized by Bitcoin”, this phrasing is a reference to dampening Bitcoin’s inherent volatility through financial engineering and economic asset coverage.

Consequently, rather than Money Market Funds (MMF) or Fixed Income, Currencies, and Commodities (FICC) products, this analysis uses Bitcoin as the primary risk-transmission benchmark, alongside STRC’s stated near-par objective. When Strategy highlights “money market like” stability or reduced volatility, they are describing compressed, realized outcomes relative to Bitcoin over time; not over specific timeframes nor as a guarantee of absolute, zero volatility MMF equivalence under all market conditions. Expecting the latter misinterprets the primary framework and subjects the product to a (likely unattainable) absolute standard. The near-par objective can still be assessed in its own right, while recognizing the broader Bitcoin-linked risk framework.

Looking at STRC through 9-4-2026, the VWAP chart shows just how far price has recovered from the breakdown and the all-time low intraday price of $71.25 on 6-26. STRC closed at $97.75 on 9-4, higher on six rolling VWAP windows, with the 5-session VWAP sitting at $97.29 and the 30-session VWAP at $94.79. Across 278 daily closes (from 7-30-2025 to 9-4-2026) STRC recorded 35 days closed at or above par, or [12.6%] of trading sessions; with another 190 sessions [68.3%] closing between $95-$99.99. In total, in 225 of 278 sessions [80.9%] STRC closed at $95 or higher. The June breakdown was an exception not the expectation over the life of the product; these are important data points to consider when looking at STRC and assessing its performance since inception.

STRC Trading Ranges as of September 4, 2026

Text version of this table

Latest Close	$97.75	-2.25% vs $100
Range Since Inception	$71.25 - $100.42	Low Jun 26, 2026 | High Jan 13, 2026
Days At or Above Par	35 of 278	12.6% | Close >= $100.00
Days Closed $95-$99.99	190 of 278	68.3% | $95.00 <= Close < $100.00
Days Closed $90-$94.99	22 of 278	7.9% | $90.00 <= Close < $95.00
Days Closed $85-$89.99	25 of 278	9.0% | $85.00 <= Close < $90.00
Days Closed $80-$84.99	4 of 278	1.4% | $80.00 <= Close < $85.00
Days Closed Under $80	2 of 278	0.7% | Close < $80.00
Latest Trading Notional	$129.4M	1,321,314 Shares
5-Day | 30-Day VWAP	$97.29 | $94.79	Rolling Volume-Weighted Daily Bars

Important Context: The June 29th Window

From Friday June 26’s close to Monday June 29’s close, STRC gained [12.20%] rising from $74.57 to $83.67 and MSTR gained [12.60%] rising from $82.31 to $92.68. On June 29, Strategy announced its repurchase authorization, an increase in STRC’s annualized dividend rate to [12%] and additional liquidity measures. The first disclosed purchases occurred during July 20-26 and were announced July 27 but the rebound actually preceded execution. In a subsequent disclosure on September 8, the Digital Credit Securities authorization was increased to $2 Billion, with approximately $1.19B remaining as of September 7. The program remains discretionary and that later authorization update is useful context; the market-data cutoff for this analysis remains September 4.

Understanding STRC Price Variance

When trying to analyze price performance on a particular equity, we have to distinguish between claims to causality (factor “A” is what moves | moved price) and statements about attributable variance (factor “A” was identified as a variable associated with a measured share of the observed variation in returns). Caution also needs to be taken when: a} Identifying what is readily observable in the data b} What can be observed in the data with proper controls c} What can be inferred from the data and d} What cannot be observed in the data or would require unsupported assumptions that contaminate results.

Understanding this, to properly assess the variance in STRC daily returns, the following Multivariate Order-Independent Decompositional Analysis (M.O.D.A) separates observed relationships in STRC into five factors in three primary categories, plus two auxiliary categories for added market context. The three primary categories are:

The 60-session model relies on complete, matched observations and eliminates multicollinearity in a specific sequence. While the construction of model inputs adheres to this sequence, the subsequent application of Shapley Attribution remains order-independent across those inputs. An auxiliary category: [4] Structural Features & Corporate Actions: Including additions to USD reserves, changes to dividend policy | frequency and share repurchases - provides additional context and (where separately tested) supplemental evidence. These features are not separately estimated factors in the primary 60-session model. Another auxiliary category; [5] Derivatives and Leverage: Including brokerage margin, borrowing against tokenized wrappers | lending loops, and traditional | on-chain derivatives - provides context for how financing terms, collateral requirements and liquidations may amplify price moves. These channels are not separately measured or attributed by the primary model and their potential role is considered in the “Market Read” section of this document. The resulting allocations describe associations with price return variance; they do not establish causal price impact. The following section details the results from the analysis on STRC.

M.O.D.A Results

Over the execution portion of the observed window (July 20 onward), Strategy’s intervention was a significant portion of trading volume. Nevertheless, volume | daily short volume flow factors ranked fourth or fifth among the five displayed groups in rolling 60-session attribution models ending during the buyback execution period. As of 9-4, the exact results from the analysis of that same 60-session (trading day) variance attribution ( 6-11 to 9-4) showed the following percentages allocated to price variance:

The 60-session orthogonal model explained ~74.65% of STRC daily-return variance within the 6-11 to 9-4 window (adjusted R² 71.24%) and the percentages above are shares of that explained variance, not shares of all STRC price variance. In simple terms out of ~74.65% of explained price variance within that window, the model allocated [~33.4%] to the other preferreds, [~29.2%] to MSTR (excluding-BTC), [~11.7%] to BTC, [~0.32%] to trading volume, and [~0.04%] to daily short-volume flow, with [~25.3%] of price variance left unexplained.

[Figure S1] STRC | 60-Session Stability Benchmark: Rolling Attribution

Buybacks contribute to trading volume and so should sit inside volume as a variable, though an unsigned, residualized volume factor is a low-power place to find a directional bid; the direct proxy in STRC Buybacks as of 9-4-2026 is the primary test. In this window, factors tied to trading volume and daily short volume flow are in the model, and their coefficients were not statistically distinguishable from zero (HAC p = 0.37 | 0.73 respectively) in the results. In contrast the preferred block, MSTR ex-BTC, and BTC factors were (HAC p < 0.001) indicating a highly statistically significant association.

The preferred block result is a joint test of three components, representing four peer offerings, while the BTC and MSTR results are tests of individual coefficients. The two factors tied to volume enter the model residualized against the [1] Asset | Parent and [2] Preferred-Market factors, and the factors themselves are orthogonalized (Max VIF {Variance Inflation Factor} of 1.00), so the 44.7 | 39.2 | 15.7 split is incremental rather than three ways of counting the same BTC related factor.

Read plainly, the largest single share of STRC price variance over this window was associated with the other preferred offerings and SATA as a block, followed by a repricing of MSTR (independent of Bitcoin), and then Bitcoin itself. It is important to note that earlier in the year, this same model was Bitcoin-dominated; by summer it became Capital Structure-dominated. Over the observed period, this shows that an instrument engineered to hold near par does not stop the market from expressing a view; it may simply relocate the view into the structure’s own credit and equity factors; with STRC priced as a claim on that structure. The results of this model add support for a perspective that the recovery occurred within a broader market repricing, rather than solely from a corporate policy driven rebound.

STRC | SATA Comparative Results

The following results [Figures S2 to S5] “look inside” the preferred market relationship for STRC and SATA. The orthogonal benchmark continues the M.O.D.A framework: overlapping co-movements are separated, then peer group contributions among its four offerings are allocated. As an additional model, the raw-factor audit enters each offering separately and shares overlapping information among the original inputs.

[Figure S2] STRC | 60-Session Benchmark: Factor Contributions & Peer Allocations

June 11–September 4, 2026 | 60 Matched Sessions | Model R² 74.65% | Adjusted R² 71.24%

Of the [~44.70%] allocated to STRC’s peer preferred block, SATA receives [21.62%] of that block; this translates to [9.66%] of the whole model’s explained variance, or [7.21%] of total STRC price-return variance. MSTR-ex-BTC remains larger than any individual peer allocation in this model. Additional context for the existing results: SATA participates in the association, alongside the Strategy preferreds but because the ticker amounts are allocated from shared components, the [9.66%] should not be ascribed to SATA’s independently identified effect on STRC.

[Figure S3] SATA | 60-Session Benchmark: Factor Contributions & Peer Allocations

June 11–September 4, 2026 | 60 Matched Sessions | Model R² 72.67% | Adjusted R² 68.99%

With SATA, the Strategy preferred block accounts for [66.68%] of SATA’s explained variance with STRC receiving [25.51%] of that block, equivalent to [17.01%] of the whole model’s explained variance and [12.36%] of total SATA price return variance. Interesting note: STRK | STRD each receive larger individual allocations than STRC in this model over this window. Read against the STRC result, the component allocation to STRC within SATA is larger than the allocation to SATA within STRC; which supports a descriptive asymmetry under this model; while the distribution across several Strategy preferreds points to a broader preferred-market relationship. This does not identify which offering moved first.

[Figure S4] STRC | 60-Session Benchmark: Raw-Factor Audit

June 11–September 4, 2026 | 60 Matched Sessions | Model R² 74.66% | Adjusted R² 70.68%

In the STRC Raw Factor Audit, SATA exists as its own raw Shapley factor, receiving [19.54%] of STRC’s explained variance, equal to [14.59%] of total STRC price return variance. Over this window, STRF has the largest raw allocation at [28.72%], followed by SATA (noted above), STRK [18.58%] and MSTR [14.14%]. Overall, model fit is almost unchanged from the orthogonal benchmark, yet the allocations change materially because shared information is distributed differently. This shows that the assumptions drive the analysis and the treatment of shared variation matters; here the raw SATA share “includes” information it shares with the other inputs; but results are not the extra “fit” from adding SATA after every other input is already present. BTC receives [15.65%] of explained variance under the BTC-first orthogonal specification and [4.28%] under the raw Shapley specification. The Raw Shapley averages the added explanatory contribution from BTC across every ordering, allocating both unique and shared information; however, note that these are different model allocations, not attribution bounds. Under this same framework, any subsequent input in the orthogonal benchmark receives the remaining variance that preceding inputs have not claimed.

[Figure S5] SATA | 60-Session Benchmark: Raw-Factor Audit

June 11–September 4, 2026 | 60 Matched Sessions | Model R² 72.81% | Adjusted R² 68.54%

In the Raw Factor Audit for SATA, STRC receives [21.85%] of SATA’s raw model explained variance, or [15.91%] of total daily SATA price return variance. Interestingly, here STRD has the largest allocation at [39.54%], followed by STRC [21.85%] and STRK [16.96%]. Overall, the 60-session reciprocal raw comparison is therefore [19.54%] for SATA within STRC against [21.85%] for STRC within SATA, with STRC modestly larger on both reported scales. In aggregate, these results are descriptive evidence of an interconnected Digital Credit | Preferred complex. They cannot speak to capital diversion, so they neither support nor refute the view that SATA is “eating” STRC; what they do show is that STRC and SATA co-moved with one another, consistent with a market of Bitcoin-correlated equities.

Cross Asset Results | A Comparative Check

The next question to assess is whether STRC was expressing a market view that was unique to STRC, or participating in a repricing that extended across Digital Credit as a block. To examine that question, the same 60-session framework was applied to STRK | STRF | STRD and SATA over June 11 through September 4, with STRC retained as a reference. Each offering is measured against BTC, its own parent equity excluding BTC-related movement, the other four preferreds, and its own trading volume | daily short-volume flow.

Preferred Block Results

When reading across the offerings, there is a commonality that is clear: the other-preferred block is the largest single displayed allocation in all five models, ranging from [44.70%] for STRC to [66.68%] for SATA. The cohort shared a strong relationship, while the relative expressions of BTC | Parent Equity | Other Preferreds differed by instrument. This adds evidence for a broader Digital Credit repricing; it does not, on its own, tell us which offering led it. The individual results add context:

[1] STRK (Strike): Out of the total Model R² of [80.44%] the convertible preferred offering carries the largest BTC share of the five at [30.96%], alongside [22.02%] MSTR-ex-BTC and [46.80%] attributable to other preferreds. Its MSTR-ex-BTC coefficient is [0.332], greater than STRF’s [0.206] and STRD at [0.231], although below STRC at [0.357]. While there seems to be some differentiation in response magnitude consistent with the equity-linked claim, there is not a simple contractual ranking in the attribution shares, nor should one be expected. The stated conversion rate from STRK is 0.1 MSTR share per preferred share, equivalent to a $1,000 conversion price per MSTR share, subject to contractual adjustments. At the June 29 close of $92.68 for MSTR, immediate conversion value was only $9.27 per STRK share. This is a view that theoretically supports emphasizing the stated 8% dividend rate and credit characteristics at those prices, while its conversion option still carries exposure to future MSTR prices.

[2] STRF (Strife): Out of a total Model R² of [66.95%] the senior, fixed-rate offering has [50.43%] allocated to the preferred block, [30.15%] to MSTR-ex-BTC and [19.35%] to BTC. Its BTC and parent coefficients are the smallest among the four Strategy offerings in this window. Note: This is consistent with a more muted response magnitude; it does not mean that seniority removes the relationship with the issuer nor does it establish that seniority caused the differences.

[3] STRD (Stride): Out of a total Model R² of [81.51%], the junior, non-cumulative offering has [59.76%] allocated to the preferred block, [28.59%] to BTC and [10.02%] to MSTR-ex-BTC with the preferred block carrying a materially greater share in STRD than it does in STRC. Daily short-volume flow as a factor is [1.62%], the largest short-flow allocation in this cohort window, although its coefficient is not statistically distinguishable from zero (HAC p = 0.096).

[4] SATA: Informative comparisons extend beyond Strategy’s preferred peer block. SATA is a senior preferred offering with a different parent equity in Strive (ASST), and despite that separation, out of a total Model R² of [72.67%] the Strategy preferred block accounts for [66.68%] of explained variance, alongside [19.12%] for BTC and [12.79%] for ASST-ex-BTC. Trading volume contributes [1.26%] which is small but statistically distinguishable from zero over the observed window (HAC p = 0.029). It is possible for a small variance allocation and statistical significance to coexist and this is additional evidence for the asset specific manifestation of each factor.

As with the rebound in STRC, assessing the timing of price rebounds across the preferred block can provide important context. Over the seven weekly repurchase disclosures that cover July 20 through September 4 there were 8,541,087 STRC shares purchased and no STRK | STRF | STRD purchases. With no repurchases, all five offerings rebounded substantially between the common June 26 and July 17 window, the final session before the first disclosed repurchase window. Afterwards, their recovery paths differed:

A Look Into Price Recovery

Text version of this table

Preferred Recovery Comparison
June 26–September 4, 2026 | Daily Closing Prices

Offering	June 26 Close	Price Return		Share of June 26 Discount to $100 Closed
		June 26–July 17	July 17–September 4	By July 17	By September 4
STRC	$74.57	14.38%	14.61%	42.2%	91.2%
STRK	$53.79	14.76%	25.98%	17.2%	51.9%
STRF	$84.37	13.91%	8.52%	75.1%	127.5%
STRD	$53.48	12.21%	26.48%	14.0%	48.2%
SATA	$87.75	11.37%	2.33%	81.5%	100.1%

Same issuer offerings will share corporate announcements and potential balance-sheet spillovers, whereas SATA has a different issuer. As a cohort, they serve as useful comparators, even though they are not a clean control group. Because the June 29 authorization also preceded these purchases, no formal difference-in-differences or causal effects are estimated; therefore no statement can be made as to what STRC would have done without buybacks. Although STRC appears in every sibling peer basket, it accounts for only [14.42%] – [29.50%] of first-component squared loadings across the four sibling models, so STRC does not dominate the first retained component in any of these four models. Those shares are from the first retained peer component in each model under the Appendix C model construction. However, because the models rely on same day returns and overlapping peer sets, directionality is also unresolved and cannot determine whether the block follows STRC or vice versa.

While the results do not establish that buybacks made no difference, this strengthens the case that the recovery in STRC was part of a broader repricing already underway before share repurchase executions began. Moreover, considering that the other market preferred offerings like STRF closed through par, and SATA closed its June 26 discount with no repurchase bid ([127.5%] and [100.1%] respectively); this shows that repurchase execution was not necessary for the recovery in those offerings. It does not establish what STRC would have done without an authorization, that counterfactual is not part of this analysis; but it does raise the question of whether or not Strategy’s intervention was necessary.

Trading | Liquidity Dynamics

With a broader market lens, it’s clear that shorting STRC is more than just a view that Bitcoin will fall in price; the common stock is arguably the most efficient vehicle for that position. Shorting around STRC can express the view that the capital structure will weaken, although shorts can also hedge or provide added liquidity. Interestingly enough, the dynamic we have seen play out recently around STRC has been four-fold: 1] Short sellers facing dividend replacement payments at the stated annualized rate, plus borrowing costs, 2] at the same time, a significant buyer over the summer has been the offering issuer buying below par. 3] Investors looking for significant gains in the form of capital appreciation from effective yield in tandem with dividend yield. 4] Another investor cohort likely seeking an exit from the offering due to increased near term volatility and increased perceived risk. The data in aggregate cannot easily delineate which cohort had what effects and in what capacity.

Due to the price activity around STRC, investors have been searching for an explanation, but most of the recent discourse has focused on distinct variables that have been somewhat conflated. Firstly, a] Short Interest is the stock of open short positions as of a settlement date: a persistent open position that must eventually be covered. While b] Daily Short Volume is a gross flow metric of reported sales marked short and can reflect market-maker activity that is covered within a session. The change in short interest equals shorting minus covering, with covering itself remaining unobserved in the data. Reported short volume and changes in short interest provide complementary context, but their different coverage and timing mean that these data cannot quantify short covering.

The newest factor to consider within trading activity are c] Share Repurchases, which reduce outstanding claims and improve the capital structure, although subsequent issuance can (and likely will) offset any near term reductions. Across the July 20 through September 4 sessions (including the repurchases announced September 8), STRC traded 48.99M shares, roughly $4.60B in total volume. The company repurchased 8,541,087 shares representing 17.43% of share volume for ~$811.5M or 17.63% in dollar volume. Over that same period the total daily short volume was 8,309,496 shares representing 16.96% of share volume or $778.33M of estimated notional value or 16.91% of dollar volume. This highlights how much of the market conversation has focused on a], paid less attention to the scale of b], and treated c] as a one-sided lever for price recovery.

Over this window, Strategy’s intervention was a significant portion of observed trading volume; nevertheless, reported short sale flows were roughly equivalent in scale. As comparative points, neither establishes net price impact since short volume flow and share repurchases cannot be netted out as if they were independent from one another; a repurchase can meet a short-marked sale in the same trade. A repurchase removes an outstanding claim, while a short sale may be covered within the session or remain open. Taken in aggregate however, this does weaken claims that the recovery toward par can be viewed as a one-sided lift from corporate action.

Quick Summary: Across the 7-20-2026 to 9-4-2026 trading sessions (Including buybacks announced 9-8-2026)

STRC Buybacks as of 9-4-2026

The claim made in the “Understanding STRC Price Variance” section is advanced here against a direct buyback proxy, as opposed to volume alone: the tested factors related to trading volume | short volume flow had small conditional associations with return variance, and a separate buyback proxy adds little explanatory fit under the tested construction. (Note: Buybacks were not a separate factor item in the displayed 60-session model).

When disclosed weekly repurchase dollar amounts are evenly allocated by session (day) as a proxy, they account for [1.30%] of total daily price-return variance if the other factors are ignored, and are not statistically distinguishable from zero once BTC, MSTR, and the other preferred offerings are in the model (HAC p = 0.50). The incremental fit for the buyback proxy after those controls is an additional 0.106 percentage points of R² in the model (ordinary R² from [74.29%] to [74.40%]; with the activity factors also included, from [74.65%] to [74.66%], HAC p = 0.83). These findings are consistent with trading volume and short flow representing 0.32% and 0.04% of total variance respectively.

In the supporting 30-session activity-attribution model (July 27–September 4; Appendices D–E), the window begins the Monday after the first disclosed repurchase week rather than on July 20, where the execution window in the sensitivity tables begins. Activity factors rise to [3.73%] and [3.00%] of explained variance, [1.47%] and [1.18%] of total variance, inside a weaker fit (ordinary R² [39.44%]; adjusted R² [20.17%]). This represents a larger share of a smaller explained slice, consistent with an instrument compressing into a band; this should not be read as a refutation of the 60-session result. The window’s residuals also show serial dependence (LB5 p = 0.014, Appendix H), so it is best read as a supporting view rather than an independent confirmation.

Referring back to the cautions presented at the start of the understanding STRC variance section, namely c} What can be inferred from the data; and the logical inference made at the start of the results section, the following interpretation is raised for consideration: ( i ) The recovery in STRC began before repurchase executions ( ii ) The other Strategy Preferred Offerings (and SATA) also recovered before repurchases occurred | without share repurchases ( iii ) Trading Volume | Short Volume Flow are in the model, and their coefficients are not statistically distinguishable from zero over the observed window ( iv ) Share Buybacks as a proxy (when evenly allocated by session) are not statistically distinguishable from zero once BTC, MSTR, and the other preferred offerings are in the model. Given ( i - iv) above, while the results of this analysis do not claim that the effect of buybacks was zero, there is evidence that they were not the main explanation for the STRC price rebound over this observed window.

Text version of the chart labels

Strategy | STRC Buybacks and Trading Volume

Reporting Period	STRC Total Trading Volume (Share Volume)	STRC Buybacks (% of Weekly Share Volume)
07/20 to 07/26	5.28M	5.47%
07/27 to 08/02	6.55M	13.92%
08/03 to 08/09	7.88M	14.61%
08/10 to 08/16	5.89M	23.58%
08/17 to 08/23	7.06M	20.28%
08/24 to 08/30	8.23M	18.93%
08/31 to 09/07	8.10M	22.35%

BTC | MSTR Sensitivity Across the Windows

The market uses Strategy, and other treasury companies, to forward-price its expectations on Bitcoin. Firstly, within the capital structure (MSTR | STRK | STRF | STRD | STRC | Convertible Notes) there is a single forward distribution of Bitcoin outcomes across claims with different payoffs, each (in theory) representing a different aspect of that distribution. Then through the derivatives market, the common stock (and various Strategy-dependent ETFs) offer further differentiated exposure through one of the most liquid single-stock options markets. The common stock prices the center and the right tail of a distribution, with the options market and various ETFs pricing the left and right tail, depending on exposure and outcomes. The preferred offerings arguably price the left tail of the distribution in “credit” terms. Along this spread of differentiated outcomes, STRC tracks both Bitcoin and MSTR, sharing their direction at a fraction of their amplitude. This is the interpretive | conceptual framework that guides this analysis and should not be viewed as a measurement of investor expectations. The daily-return comparisons below align with this framework and help show how STRC’s relationship with BTC and MSTR changed through the selloff and recovery.

Transmission Channel Sensitivity

May 28–June 26, 2026 | Selloff [21 Trading Days]

During the selloff, STRC daily returns tracked MSTR more closely than BTC directly.

May 28–July 17, 2026 | Through the Pre-Buyback Rebound [35 Trading Days]

When the early price rebound in STRC is included, the strength of the MSTR relationship is more evident, while there is a weaker correlation and lower explanatory fit with BTC. This is consistent with an issuer | capital-structure interpretation, rather than STRC simply tracking BTC.

July 20–September 4, 2026 | Buyback Execution Period [35 Trading Days]

Over this period, sensitivity toward both BTC and MSTR decreased, though MSTR maintained a stronger daily return co-movement. This shift does not imply that the capital structure’s fulcrum (BTC | MSTR) lost significance; rather, daily co-movement dropped as STRC narrowed into a tighter price band while BTC and MSTR continued to fluctuate. STRC’s daily return volatility also fell from 3.11% over May 28–July 17 to 1.17% over July 20–September 4. Together, these observations describe lower realized volatility, a narrower price band, and evolving transmission channel sensitivity; this comparison alone does not establish what caused the compression, or attribute compression to share repurchases.

Market Read | As of September 4th, 2026

When the issuer of a preferred equity (in this case Strategy) signals that it is willing to use buybacks to intervene in a market it created, that changes how market participants trade the offering. Intervention might invite probing of the levels the issuer is prepared to “defend” more readily than it invites a reversion to par. A known buyer gives participants impetus to countertrade the intervention; participants sell into anticipated issuer demand to test how persistent the buying is and where it steps away. In that dynamic, the market is trading the issuer’s bid as well as STRC itself.

An alternative interpretation (based on the results from the analysis) is that the buybacks may be serving as an offramp for investors exiting STRC after discovering that its volatility exceeded their expectations: a situation where the investors are, ironically, using the issuer as exit liquidity. Irrespective of which view one holds, the issuer’s known, discretionary bid represented ~17% of share volume, while the session allocated buyback proxy added little explanatory fit after the return controls. There are many potential interpretations of this, including participants (short sellers and investors alike) selling into inorganic demand.

What the buyback debate has largely talked past (rather than about) is why STRC needed an offramp in the first place, and there is converging circumstantial evidence that points at leverage; leverage that came with the instrument’s scale and its kind. Proximity to Bitcoin is why STRC (and SATA) carry the highest stated dividend rates among the five offerings: the rate is compensation for risk in the capital structure that flows from BTC and the respective issuer; and the path of that rate from 9% at launch to 12% as of September 4th shows that risk premium being discovered in real time. An offering with a 12% coupon does not attract the investor base of a bank issued preferred like JPM-PC or WFC-PL. These “Bitcoin backed” perpetual preferreds naturally select from a different cohort, one native to (and familiar with) Bitcoin risk, yield-seeking, and leverage.

As this cohort grew in scale (~$8.5B in notional outstanding in nine months by 5-3-2026 with ~80% retail ownership by management’s estimate), leverage products naturally followed because the ability to scale is what justified offering them. There would be little incentive to tokenize, or list perpetuals on, a preferred a fraction of STRC’s size, evidenced by the fact that the adjacent instruments (STRF | STRK | STRD) and competing offerings (SATA), which offer similar premiums at a fraction of the scale, have not attracted this level of development. What financial engineering towards a par anchor added was the potential for a multiplier; compressing volatility presumably raised the loan to value that margin desks and on-chain markets would extend against the shares. In a way, the same feature that priced the risk into the dividend also shielded that risk from collateral desks.

Leverage on STRC was essentially a mispriced position on the gap between compensated risk and measured (or assumed) risk, until the two measures inevitably collided. Leverage likely accumulated through three venues: [1] Traditional Brokerage Margin Requirements [2] Borrowing against Tokenized Wrappers, such as Apyx’s apyUSD used as collateral on Morpho; [3] Traditional | On-Chain Derivatives: listed options and a June 23 vendor announcement advertising STRC perpetuals with up to 10x leverage through Hyperliquid; three sessions before the low. The very design that “stripped away” volatility created the conditions for leverage to reintroduce volatility via different channels. The availability of leverage on STRC offers a plausible explanation for the speed and depth of the selloff, considering during that period, STRC’s fitted daily-return sensitivity was roughly 3.7 times its pre-stress level to MSTR and 2.9 times its pre-stress level to BTC.

The responses from Strategy to date have focused on [4] Structural Features & Corporate Actions through increasing the USD Reserve, and launching its Digital Credit Securities Repurchase Program. However, none of the corporate actions can effectively address the reality of leverage. Arguably, STRC is at the mercy of its own success: the less volatile it becomes, the more leverage the next cohort of speculators will be permitted to apply. Investors should be cognizant of the fact that buybacks retire a claim in the form of dividend obligations, and benefit the capital structure; which is a real tangible benefit. Nevertheless, there is not strong evidence that the buybacks were the main explanation for the move STRC has made closer to par. Given the considerations of leverage, ongoing cost, opportunity-cost and market optics, the open question is therefore not whether Strategy can keep buying, it may be whether or not they should.

The conditions to watch for around a reversion to par, and the ability to maintain par with sustained independent demand, are improvements in the market framework for STRC, meaning: [1] Asset | Parent Transmission Channels: Represented by BTC and MSTR (the fulcrum of the capital structure) stabilize or strengthen and the risk premium attached to the structure either diminishes or is offset by price appreciation. [2] Preferred-Market Transmission Channels: STRK | STRF | STRD and SATA remain the largest associated factor in STRC’s variance, and a more stable complex can pull with STRC rather than against it. [3] Market Activity: trading volume normalizes and organic (unlevered) demand re-establishes itself. [4] Structural Features & Corporate Actions are reduced and allow the market to find its own equilibrium, and factors tied to [5] Derivatives and Leverage can either be absorbed due to STRC and the Digital Credit Complex being more stable due to [1] - [4] or be mitigated through adaptations.

Finally, [6] the thesis of this analysis is falsifiable on its own terms: if STRC regains par while the preferred block and MSTR ex-BTC remain weak, the capital-structure reading would be weakened; if STRC fails to hold par while they strengthen, the reading that STRC is mostly impacted by the preferred block would be weakened and a STRC-specific culprit or residual (leverage, Strategy bid-dependence, or both) remains live. The first stress window after par is regained can be viewed as a test of this market read. A return toward this summer’s stress betas of [0.34] to [0.43] versus MSTR would be consistent with a return of stronger daily return sensitivity. A beta that stays compressed near the execution window [0.098]; or near the January 2–May 27 pre-stress level of [0.092] (noted in Appendix B) would be consistent with either a changed holder base or continued band compression, but would not by itself isolate those two readings. What remains to be seen is how much independent (and unlevered) demand is left once Strategy’s intervention steps back. Stepping away from the market because demand recovered reflects a different market reality than stepping away while demand is still weak.

APPENDIX

Appendix A: STRC | BTC: Simple Rolling Correlation and Beta Snapshot

Text version of this appendix

Appendix A | STRC & BTC

Rolling Correlation & Beta | Simple Daily Price Returns
As of September 4, 2026 | 15, 20 & 30 Session Rolling Windows

Sample Fit | Beta (β): 0.364 | Correlation (r): 0.393 | R²: 15.45%

May 28–September 4, 2026 | 70 Matched Return Sessions

Latest Rolling Correlation | 15 Sessions: 0.659 | 20 Sessions: 0.583 | 30 Sessions: 0.341

Fixed Windows Used in the Report

Return Dates (2026)	Sessions	Beta (β)	Correlation (r)	R²
January 2–May 27	100	0.161	0.662	43.89%
May 28–June 26	21	0.473	0.519	26.94%
May 28–July 17	35	0.508	0.425	18.06%
July 20–September 4	35	0.162	0.336	11.30%

Daily simple price returns, excluding dividends, as of September 4, 2026. Beta describes the fitted return response; correlation describes how closely the returns move together. The sample fit uses 70 matched returns from May 28–September 4. The fixed rows use 100, 21, 35 and 35 sessions. These are different samples from the latest 60-session multi-factor estimate.

January 2–May 27 is the pre-stress baseline. Rolling windows retain earlier history. May 28–June 26 and May 28–July 17 overlap. Dividends are excluded. Descriptive associations; no causal attribution.

Return Basis: BTC closes matched to equity dates; returns span intervening weekends.

Appendix B: STRC | MSTR: Simple Rolling Correlation and Beta Snapshot

Text version of this appendix

Appendix B | STRC & MSTR

Rolling Correlation & Beta | Simple Daily Price Returns
As of September 4, 2026 | 15, 20 & 30 Session Rolling Windows

Sample Fit | Beta (β): 0.261 | Correlation (r): 0.630 | R²: 39.63%

May 28–September 4, 2026 | 70 Matched Return Sessions

Latest Rolling Correlation | 15 Sessions: 0.684 | 20 Sessions: 0.635 | 30 Sessions: 0.475

Fixed Windows Used in the Report

Return Dates (2026)	Sessions	Beta (β)	Correlation (r)	R²
January 2–May 27	100	0.092	0.650	42.20%
May 28–June 26	21	0.340	0.729	53.12%
May 28–July 17	35	0.426	0.759	57.62%
July 20–September 4	35	0.098	0.471	22.23%

Beta describes the fitted STRC return response; correlation describes how closely the returns move together. These are separate two-series relationships, not the multi-factor allocations in the main paper. R² measures the share of STRC daily-return variance fitted by the benchmark alone; it is not a causal percentage.

January 2–May 27 is the pre-stress baseline. Rolling windows retain earlier history. May 28–June 26 and May 28–July 17 overlap. Dividends are excluded. Descriptive associations; no causal attribution.

Return Basis: MSTR closes on a consistent share basis; the overall relationship includes BTC exposure.

Appendix C: M.O.D.A Factor Definitions | Handling Methods

Text version of this appendix

Appendix C | M.O.D.A Factor Framework

STRC | Definitions & Handling Methods
June 11–September 4, 2026 | 60 Matched Sessions

STRC accepted model | June 11–September 4, 2026 | 60 matched sessions. The first three categories contain five displayed factor groups and seven fitted inputs because the peer group uses three components. Returns exclude dividends.

Classification	Factor or Role	Definition and Handling
[1] Asset and Parent Transmission Channels	BTC Core	Simple BTC price returns matched to equity trading dates. Retain BTC as the first modeled input.
	MSTR Ex-BTC	MSTR price returns after removing their linear relationship with BTC within each 60-session window.
[2] Preferred-Market Transmission Channels	Other Preferreds Ex-BTC/MSTR	Use STRK, STRF, STRD and SATA, excluding STRC. Remove their linear relationships with BTC and MSTR within each window, then retain the first three statistical components of the remaining peer returns.
[3] Market Activity	STRC Trading-Volume Deviation	Take log daily dollar volume (daily VWAP × shares traded) minus its prior 20-session mean log volume. Then remove linear relationships with BTC, MSTR Ex-BTC and the retained peer components.
	Reported Daily Short-Volume Flow Deviation	Subtract the prior 20-session mean short-volume percentage from the current percentage, both entered on a 0–100 scale. The percentage uses source-covered total volume. Then remove linear relationships with all preceding inputs, including trading volume.

[4] Structural Features and Corporate Actions: USD reserves, dividend policy, rate and payment frequency, and share repurchases provide context or, where separately tested, supplemental evidence. They are not separately attributed factors in this model.

[5] Derivatives and Leverage: Traditional brokerage margin, borrowing against tokenized wrappers, lending loops, and traditional or on-chain derivatives are potential amplification channels. Leverage and liquidations are not directly measured or separately attributed by this model.

Appendix D: Shapley Benchmark Fit

Text version of this appendix

Appendix D | Shapley Benchmark Fit

As of September 4, 2026

The latest 30-session window is July 27–September 4, 2026. The latest 60-session window is June 11–September 4, 2026. Each uses complete, matched trading-session observations. The shorter window sits inside the longer one; these are overlapping views, not independent confirmations.

Model Fit | Ordinary and Adjusted R²

Target	Model	Sessions	Model R²	Adjusted R²	N	Residual DF
STRC	Orthogonal	30	39.44%	20.17%	30	22
STRC	Raw	30	46.16%	25.65%	30	21
STRC	Orthogonal	60	74.65%	71.24%	60	52
STRC	Raw	60	74.66%	70.68%	60	51
SATA	Orthogonal	30	39.47%	20.21%	30	22
SATA	Raw	30	43.33%	21.74%	30	21
SATA	Orthogonal	60	72.67%	68.99%	60	52
SATA	Raw	60	72.81%	68.54%	60	51

Model R² is the percentage of the target’s total daily price-return variance explained in that window. These fits describe the observed sample; they are not validated out-of-sample predictions. Adjusted R² allows for sample size and the number of estimated inputs. The orthogonal model has seven regressors plus an intercept; the raw model has eight plus an intercept. Residual degrees of freedom are the observations remaining after those parameters are counted.

Reciprocal Allocations | Both Percentage Bases

Model	Sessions	SATA in STRC % of Explained	SATA in STRC % of Total	STRC in SATA % of Explained	STRC in SATA % of Total
Orthogonal	30	4.40%	1.73%	43.78%	17.28%
Raw	30	35.12%	16.21%	35.93%	15.57%
Orthogonal	60	9.66%	7.21%	17.01%	12.36%
Raw	60	19.54%	14.59%	21.85%	15.91%

Each direction uses a different target and therefore a different variance denominator. These percentages compare descriptive allocations; they do not measure capital transferred, the sign of a relationship, a causal price effect, or which instrument moved first. The orthogonal peer figures allocate shared components to tickers. The raw figures are Shapley allocations to original inputs.

Appendix E: STRC Factor Contributions & Peer Allocations

Text version of this appendix

Appendix E | STRC Factor Contributions & Peer Allocations

30 Sessions: July 27–September 4, 2026
60 Sessions: June 11–September 4, 2026

Every percentage below names its base: “explained” means the corresponding model’s ordinary R²; “total” means all of the target’s daily price-return variance. The explained columns sum to 100%, while the total columns sum to model R². Rounding can leave small differences.

Orthogonal Benchmark | Factor Contributions and Peer Allocations

Factor or Allocation	30 Sessions % of Explained	30 Sessions % of Total	60 Sessions % of Explained	60 Sessions % of Total
BTC Core	29.41%	11.60%	15.65%	11.69%
MSTR Ex-BTC	28.02%	11.05%	39.17%	29.24%
STRK Peer-Block Allocation	9.81%	3.87%	13.19%	9.85%
STRF Peer-Block Allocation	16.90%	6.66%	6.57%	4.91%
STRD Peer-Block Allocation	4.75%	1.87%	15.27%	11.40%
SATA Peer-Block Allocation	4.40%	1.73%	9.66%	7.21%
Dollar-Volume Residual	3.73%	1.47%	0.42%	0.32%
Short-Volume Residual	3.00%	1.18%	0.06%	0.04%

Raw-Factor Audit | Individual Input Shapley

Factor or Allocation	30 Sessions % of Explained	30 Sessions % of Total	60 Sessions % of Explained	60 Sessions % of Total
BTC Return	6.13%	2.83%	4.28%	3.19%
MSTR Return	17.65%	8.14%	14.14%	10.56%
STRK Return	13.58%	6.27%	18.58%	13.87%
STRF Return	15.33%	7.08%	28.72%	21.44%
STRD Return	6.31%	2.91%	12.65%	9.45%
SATA Return	35.12%	16.21%	19.54%	14.59%
Log Dollar-Volume Deviation	4.65%	2.14%	0.55%	0.41%
Short-Volume Share Deviation	1.24%	0.57%	1.54%	1.15%

Orthogonal peer amounts are allocated from three shared peer components using their squared loadings. Eight displayed rows therefore represent seven regressors, not eight separately estimated orthogonal factors. Raw peer amounts are exact Shapley allocations to four individual peer returns with their overlap preserved.

The ordinary-R² share is the allocation divided by ordinary model R². Adjusted R² is never used as the allocation denominator. Neither type of allocation should be read as a unique causal contribution, coefficient, or capital-flow estimate.

Appendix F: SATA Factor Contributions & Peer Allocations

Text version of this appendix

Appendix F | SATA Factor Contributions & Peer Allocations

30 Sessions: July 27–September 4, 2026
60 Sessions: June 11–September 4, 2026

Every percentage below names its base: “explained” means the corresponding model’s ordinary R²; “total” means all of the target’s daily price-return variance. The explained columns sum to 100%, while the total columns sum to model R². Rounding can leave small differences.

Orthogonal Benchmark | Factor Contributions and Peer Allocations

Factor or Allocation	30 Sessions % of Explained	30 Sessions % of Total	60 Sessions % of Explained	60 Sessions % of Total
BTC Core	7.73%	3.05%	19.12%	13.89%
ASST Ex-BTC	14.21%	5.61%	12.79%	9.29%
STRC Peer-Block Allocation	43.78%	17.28%	17.01%	12.36%
STRK Peer-Block Allocation	6.53%	2.58%	22.66%	16.47%
STRF Peer-Block Allocation	2.32%	0.91%	5.83%	4.24%
STRD Peer-Block Allocation	11.03%	4.35%	21.17%	15.39%
Dollar-Volume Residual	13.19%	5.21%	1.26%	0.91%
Short-Volume Residual	1.20%	0.47%	0.16%	0.11%

Raw-Factor Audit | Individual Input Shapley

Factor or Allocation	30 Sessions % of Explained	30 Sessions % of Total	60 Sessions % of Explained	60 Sessions % of Total
BTC Return	2.83%	1.23%	4.32%	3.15%
ASST Return	10.63%	4.61%	8.39%	6.11%
STRC Return	35.93%	15.57%	21.85%	15.91%
STRK Return	10.05%	4.35%	16.96%	12.35%
STRF Return	19.97%	8.65%	5.74%	4.18%
STRD Return	5.35%	2.32%	39.54%	28.79%
Log Dollar-Volume Deviation	14.50%	6.28%	1.66%	1.21%
Short-Volume Share Deviation	0.74%	0.32%	1.54%	1.12%

Orthogonal peer amounts are allocated from three shared peer components using their squared loadings. Eight displayed rows therefore represent seven regressors, not eight separately estimated orthogonal factors. Raw peer amounts are exact Shapley allocations to four individual peer returns with their overlap preserved.

The ordinary-R² share is the allocation divided by ordinary model R². Adjusted R² is never used as the allocation denominator. Neither type of allocation should be read as a unique causal contribution, coefficient, or capital-flow estimate.

Appendix G: Methodology & Definitions

Text version of this appendix

Appendix G | Methodology & Definitions

Returns, Model Construction, Statistical Tests & Percentage Bases

Price Returns and Sessions

The outcome is the daily simple close-to-close preferred share-price return: today’s close divided by the preceding matched close, minus one. Dividend cash is excluded. BTC levels are matched to the equity-session calendar before returns are calculated. A 30- or 60-session model contains that many matched return observations, not calendar days. Recovery returns compare the named closing prices directly; June 26 is a common anchor, not necessarily every security’s own trough. The 3.11% and 1.17% volatility figures are sample standard deviations of daily simple STRC returns for May 28–July 17 and July 20–September 4, respectively. They are not annualized.

Orthogonal Benchmark

The model keeps BTC first, then the part of the parent’s return not linearly explained by BTC. It removes both from the four peer returns and retains three peer components. It then uses the part of target trading activity not explained by those return factors, followed by the part of short-volume activity not explained by the earlier factors. This construction removes overlap in a defined sequence. Shapley averaging is order-independent across those resulting inputs; it does not remove the ordering choice that created them.

Raw-Factor Audit

The raw audit keeps BTC, parent, four peer returns, and the two target activity deviations as eight original inputs. It evaluates all 256 possible input combinations, including the empty-model baseline, and averages each input’s added explanatory fit over all possible orders. Shared information is allocated among inputs. Raw Shapley is therefore not simply the unique fit gained by adding a factor last, nor an estimate of a causal effect.

Trading and Short-Volume Activity

Dollar volume is daily vendor VWAP multiplied by reported share volume. The raw dollar-volume deviation is log dollar volume minus its average over the preceding 20 sessions, requiring at least 10 prior observations. Short-volume-share deviation is the reported short-volume percentage (on the 0–100 scale) minus its corresponding preceding-session average. Both baselines exclude the current session. The orthogonal benchmark then residualizes these deviations against the preceding model inputs.

Statistical Tests and Diagnostics

HAC means heteroskedasticity- and autocorrelation-consistent uncertainty estimates: it adjusts coefficient tests for uneven and serially related regression errors. The preferred-block result is a joint test of the three retained components representing four peers. BTC, parent and activity p-values test their own coefficients. The production lag rule gives three lags in both latest windows. These are nominal p-values, without an adjustment for multiple comparisons; a large p-value does not prove the effect is zero. The coefficient and joint tests do not test the individual ticker Shapley allocations. No Shapley confidence intervals or rank-uncertainty tests are supplied.

Reading the Percentages

A contribution expressed as a percentage of total daily price-return variance is 100 times its R² contribution. Its percentage of explained variance is 100 times that contribution divided by ordinary model R². A peer-block-only share instead divides a ticker allocation by the peer block’s contribution; this is a third, explicitly named denominator. For the dollar buyback proxy, “fit rises from 74.29% to 74.40%” gives the two model levels directly and avoids confusing a change in percentage levels with a relative percentage increase.

Supplemental Buyback Proxy

Seven weekly repurchase dollar disclosures are evenly allocated across 35 execution-period sessions, with zero allocations in the 25 earlier sessions of the 60-session window. The proxy test uses a fixed HAC lag of three, the same unscaled covariance and a Student-t reference. The proxy alone explains 1.30% of total daily price-return variance. Adding it after BTC, parent and peer controls changes ordinary R² from 74.29% to 74.40%, with HAC p = 0.50. This is the return-control specification, excluding the activity factors. A separate full-control check, including activity factors, changes R² from 74.65% to 74.66%, with HAC p = 0.83. The proxy does not recover daily executions, control for every policy channel, or establish the absence of a price effect.

Appendix H: Latest-Window Diagnostics & Source Boundaries

Text version of this appendix

Appendix H | Latest-Window Diagnostics & Source Boundaries

As of September 4, 2026

Latest-Window Diagnostics

Target	Model	Sessions	Rank	Max VIF	DW	LB5 p	LB10 p
STRC	Orthogonal	30	7	1.00	1.85	0.014	0.025
STRC	Raw	30	8	2.95	2.14	0.096	0.160
STRC	Orthogonal	60	7	1.00	2.05	0.921	0.693
STRC	Raw	60	8	5.17	2.05	0.919	0.705
SATA	Orthogonal	30	7	1.00	2.10	0.121	0.053
SATA	Raw	30	8	4.08	2.06	0.274	0.105
SATA	Orthogonal	60	7	1.00	1.77	0.334	0.334
SATA	Raw	60	8	5.15	1.75	0.366	0.391

VIF describes overlap among model inputs; the orthogonal model’s value near one is a property of its construction, not proof that the underlying market variables are unrelated. Raw overlap remains visible: the 60-session maximum VIF is 5.17 for STRC and 5.15 for SATA. Rank counts the independent input columns; all latest designs have full rank.

Durbin–Watson (DW) and Ljung–Box (LB) examine residual serial dependence. The STRC 30-session orthogonal model has LB5 p = 0.014 and LB10 p = 0.025, so that shorter-window reading needs a residual-dependence caveat even though DW is near two. Passing these diagnostics does not demonstrate causality, eliminate model risk, or turn overlapping windows into independent evidence.

Full Screenshots and the Historical Source Boundary

The full 60-session screenshots show 137 rolling endpoints from February 20 through September 4, 2026. They preserve the saved historical calculations. Earlier SATA-model endpoints from February 20 through May 4 contain an uncorrected reverse-split adjustment in the historical ASST model input on February 6; those early movements should not be treated as validated market shifts. The latest 60-session window, June 11–September 4, and latest 30-session window, July 27–September 4, are after that boundary and are unaffected by it.

As-of Date and Disclosures

Market observations stop on September 4, 2026. The September 8 filing is used retrospectively for repurchases through the final included trading session; it also describes a later authorization change. That later action is not treated as an event already in force on September 4. Reported short volume covers the source’s reported short-marked sales and is not short interest. Its dollar figure uses daily overall VWAP and is an estimated notional, not a record of short-sale execution prices.

Source Notes

1. Strategy Capital Framework and STRC Rate, June 29, 2026
2. Strategy Initial STRC Repurchases, July 27, 2026
3. Strategy SEC Disclosure, September 8, 2026 — Subsequent disclosure; market data still end September 4.
4. STRC Launch Terms, July 25, 2025
5. Strategy Q1 2026 Earnings Presentation — Page 62: notional; pages 75–76: management/Broadridge retail estimate.
6. STRK SEC-Filed Investor Briefing, August 2026 — Conversion terms and option time value.
7. Strive SEC Disclosure, August 13, 2026 — SATA’s stated rate for periods beginning September 1.
8. Apyx Morpho Announcement, March 16, 2026 — Borrowing access; not a measure of utilization.
9. Apyx Royco Launch First-Ever Tranched Onchain Digital Credit Yield Product — Includes STRC and SATA exposure.
10. STRC Stock Futures Go Live — Strategy’s STRC Stock Futures Goes Live on Hyperliquid Amid Volatile Comeback
11. Armitage | Morpho Discussion, June 2026
12. Saturn Documentation — Current product architecture, not a historical measure of June activity.
13. FINRA Short-Sale-Volume Interpretation Notice — Coverage and differences from short interest.