TSLA Strategy Analysis

A lever-by-lever walkthrough of the covered-calls + X3 strategy decision, anchored in real historical episodes. Each chapter ends with an insight you can verify by clicking into the deep-dive report. Design rationale: see analysis/PROCESS.md.

Section 1 · Levers

What each knob buys you, what it costs, and when the trade flips. Each chapter contrasts 2–3 strategies that differ by one lever, picks the period that best illustrates the mechanism, and embeds the key chart inline with a link to the full deep-dive report.

1.1 · X3 weight (0 / 15 / 30 / 100%)

X3 sleeve = leveraged 3× TSLA exposure, MA10-gated (exits when TSLA closes below 10-day MA). Not raw 3× — the gate prevents the worst leg of bear markets.

All 4 strategies separate cleanly throughout the period. Same drawdown timing (Nov 2021 → Jan 2023); 0_100 trough slightly later (gate exits).

stockonly → 85_15: +$11M NAV for +1.5pp DD. 85_15 → 70_30: +$20M for +1.5pp. 70_30 → 0_100: +$149M for +8.1pp. Each step pays out — even the steep last one.

Why X3 wins so cleanly here:

• TSLA compounded ~50% CAGR over 7 years — leveraged exposure to a strong uptrend amplifies more than it damages.
• MA10 gating exits X3 before the worst leg of the 2022 bear (sleeve capped at −85%, not the −99% raw 3× would deliver).
• Stock and X3 sleeves both drew down in 2022, but X3's gated exit + faster recovery on the 2023 bounce more than offsets the bear pain.
• 2026-05 calibration: drag re-fitted from 15%/yr → 0%/yr against real Avanza cert (synth no-drag matches AVA 1 within 2pp over 6yr). All X3-bearing NAVs revised upward materially (0_100: $105M → $189M).

When X3 weight would flip:

• Range-bound TSLA: MA10 whipsaws (false signals) accumulate into death-by-a-thousand-cuts, eroding X3 NAV without the bull-market upside to make up for it.
• Slow grinding bear: gate exits late on each leg-down; cumulative X3 losses dominate.
• Neither regime is observed in TSLA 2019–2026 — claim is scoped to this dataset.

1.2 · Trim X3 (on / off)

Trim = move X3 above-target gains into stock; never refill X3 below target. Pitch: lock bull gains, avoid adding to falling X3.

70_30 vs 70_30_trim. Trim trails after the 2020 melt-up and never closes. Drawdown band: identical timing, identical magnitude.

Why DD protection fails:

• Trim resets X3 to target, never below. Peak X3 weight = no-trim peak X3 weight.
• Trimmed gains sit in stock, not cash. Stock fell ~70% in 2022 alongside X3.
• Net: same exposure at peak, ~same DD%, less compounding through the bull.

All 6 configs red on NAV (−22% to −44%). DD bars within ±0.5pp — noise.

When trim would win: regime where X3 drew down materially harder than stock and trimmed-into-stock survived. Neither held in TSLA 2019–2026.

1.3 · CC strike (none / 5 / 10 / 15 / 20% OTM)

On the 70_30 chassis (no CC equivalent for 0_100 — no stock to write against). Full coverage for now; partial coverage is chapter 1.4.

no-CC (top right): max NAV, worst DD. cc05 (bottom left): least NAV, best DD. cc10 dominated by cc05 (same NAV, worse DD). cc15 close to dominated by cc20.

Mechanism:

• CCs cap upside at strike → in TSLA's bull years, ~22–44% of cycles get assigned, locking gains at strike vs the much higher run.
• Premium income is real ($11–29M total over 7yr) but doesn't compensate for the foregone upside in a strong-trending stock.
• DD relief comes from premium cushion + reduced exposure on the way down (assigned shares are out before the bottom).
• Lower strike (cc05) → more premium, more assignment, more DD cushion, more upside cap. Higher strike (cc20) → less of all of that.

When CCs would be the right call:

• Sideways/range-bound regime: premium accrues without assignment, no upside foregone.
• If the cashflow floor matters MORE than NAV maximization (covered in 1.4 and 1.6).
• If the +8pp DD relief is psychologically valuable enough to justify ~37% NAV cost.

1.4 · Coverage mode (full vs partial-target)

Full = sell CCs against 100% of stock sleeve. Partial (target3k) = size CCs to hit ~$3K of premium per cycle (just enough for the cashflow floor), leaving the rest of stock uncovered.

Vertical right cluster: no-CC + partial variants share the −77% DD wall. Diagonal left cluster: full coverage trades NAV for DD relief.

Mechanism:

• Partial coverage sizes contracts to ~$3K of premium → typically <5% of stock sleeve covered → ~95% of upside retained.
• Premium cushion in 2022 bear: full cc05 = $29M vs partial = $0.3M. Partial has nothing to soften the drawdown.
• Same strike, same per-contract assignment rate (~44% at cc05) — partial just writes far fewer contracts.
• Among partial variants: cc05_target3k beats cc20_target3k by ~$2M of NAV — closer-to-money calls earn more per contract, so fewer contracts needed for $3K target → less upside given up.

1.5 · Puts (on / off, strike) — biggest finding

Puts alone cost premium with no income. Combined with full CCs (collar), CC premium funds the put. Tested: put15 (15% OTM put) and put20 (20% OTM put) on the 70/30 chassis.

Collar (top-left) is the only strategy on the upper-left frontier — best NAV AND best DD. Puts alone (bottom-right) and CC alone (bottom-middle) both worse than no-CC. The combo is synergistic, not additive.

Through 2019–2021 the collar tracks below no-CC (paying for puts). In the 2022 bear band, the collar holds while no-CC craters. By 2024+, collar matches/exceeds no-CC because more capital survived to compound.

Why the synergy works (mechanism):

• CC premium ($48M+ over the period at cc05_put15) more than funds the put protection cost ($17M).
• In the 2022 bear, the put hard-floors the downside around −15% from spot per cycle → MaxDD compresses from −77% to −56%.
• More capital survives the bear → more compounding base for the 2023 recovery → collar's NAV catches up and exceeds no-CC by 2024.
• The CC's upside cap matters less when the put has prevented capital destruction earlier.

Why puts ALONE fail (without CC):

• 70_30_put15 pays put premium with no income to offset → NAV drag of −38% vs no-CC, only +1.5pp DD relief.
• Without CC funding, puts are an unsubsidized insurance cost.

Put strike sensitivity (added after first pass):

Initial test showed cc05_put15 wins; testing tighter and wider puts revealed the trend continues monotonically. Tighter put = better on every metric.

Default variant. MaxDD here is the deeper of (early-2019 startup noise) or (2022 bear); the 2022 trough column is the comparable peak-to-trough in dollars.

Monotonic: tighter put = both higher NAV AND less negative MaxDD. cc05_put05 sits in upper-left (best on both axes).

Mechanism (why tighter puts win):

• Put cost is ~CC premium across the period (cc05_put05: CC $108M, put cost $106M, near zero net).
• Tight puts paid out tens of $M in 2022 → preserved capital → bigger compounding base for the 2023+ recovery.
• The "expense" of tighter puts is recovered (and then some) through preserved compounding.
• If TSLA had grinded sideways/up without the 2022 crash, tighter puts would have been pure drag — the strategy is regime-dependent on having at least one major drawdown to pay them back.

Caveats:

• Synthetic pricing — mostly N/A. Real-options chain covers Jan 2019 → Apr 2025 (~86% of cycles). Synthetic kicks in only for the post-Apr-2025 tail and uses the calibrated σ model (median +5% bias vs market, vs legacy +19%). Pure-synthetic gap on the collar would be ~24-34% NAV overstatement on legacy 1.12 — but this only shows up in the ~14% synthetic tail of the data, so end-state NAV impact is <1%.
• Regime dependence. Tight puts win because TSLA had a major bear in 2022. Without a crash in the next 7 years, looser puts (or no put) would win. Yusuf's "live with this for years" framing means betting on whether crashes recur — historically TSLA has had multiple −50%+ drawdowns, so betting on more crashes is reasonable.
• Collar requires the full CC sleeve. Doesn't combine with partial-coverage (target3k) — that has too little CC premium to fund the put.
• 0_100 has no collar option. No stock to write CCs against → no funding for puts → the all-X3 chassis can't access this winning strategy class.

1.6 · Cashflow target ($3K / none)

_w3kusd = withdraw $3K cash per cycle (~$12K/mo at the monthly cycle cadence). Funded from CC premium first, then stock sales if needed. ~$250K cumulative withdrawn over the 7yr period.

Withdrawal cost is roughly proportional. Collar's $9M edge over no-CC is preserved (~$44M vs $35M with $3K/cycle drained).

Mechanism:

• $250K cumulative withdrawn → forgone compounding at ~70–83% CAGR → that compounded delta IS the ~12–14% NAV cost.
• Withdrawal funds itself from cash/premium; no forced asset sales at bad times → DD unchanged.
• Floor sustainability: every strategy ended above $7M with the floor active. The trough during 2022 was a fraction of peak but never approached zero on any strategy.

Important non-result: partial coverage + put + withdrawal (70_30_cc05_target3k_put15_w3kusd) = $22M, the worst performer. Partial coverage's tiny premium ($0.3M) can't fund the put cost; the combination gets the worst of all worlds. If you want puts, use full coverage.

Section 2 · Shortlist

Head-to-head comparisons of the strategies that survive Section 1 as plausible candidates. Grows over time as new ideas emerge from lever insights.

Survivors of Section 1 + cross-lever search

Ruled out by lever analysis:

• Trim (1.2) — dominated across all 36 datapoints.
• stockonly, 85_15 (1.1) — dominated by 70_30 on every metric.
• CC alone, puts alone (1.5) — dominated by collar.
• Partial coverage (1.4, 1.6) — dominated by no-CC + auto-stock-sale withdrawal.
• Wider puts (put15→put20), wider CC strikes (cc10/cc15) — dominated by tighter strikes (1.5).

Cross-lever search — added during this section:

• Section 1 only tested levers in isolation on the 70/30 chassis. The collar finding (1.5) plus the X3-weight finding (1.1) suggested a cross-lever sweet spot might exist: more X3 + tight collar to control the bigger drawdowns.
• Tested: 50_50_cc05_put10_w3kusd, 50_50_cc05_put05_w3kusd, 30_70_cc05_put05_w3kusd.
• Result: 50_50_cc05_put05_w3kusd beats 0_100 on every metric (12% more NAV, 29pp better DD, lower trough wealth slightly but much smaller DD%, higher Sharpe, auto-cashflow). 0_100 is now dominated.
• 30_70 overshoots — too little stock to fund the collar economically; DD blows out without enough NAV gain to justify.

Final 6 candidates (all with $3K/cycle floor active where the model permits):

4-panel scan of the finalists. Tight collar wins NAV (among non-0_100), MaxDD%, Sharpe. Trough panel needs the * caveat above.

Final NAV across all 6 timing variants. 50/50 tight collar (Pick A): $137–320M (2.3× spread). 0_100: $79–189M (2.4× spread). Both are timing-sensitive. Conservative collar (put15) tighter at 1.9× spread.

Through 2019–2021 the four 70/30-chassis variants track tightly (collar slightly behind paying for puts). 2022 is the divergence point. Tight collar (orange) and conservative collar (green) hold while 70/30 (blue) drops with 0_100 (purple). Post-2023 the tight collar pulls ahead of all 70/30 strategies and trails only 0_100.

Pareto frontier — who dominates whom?

• 70_30_w3kusd dominated by 70_30 collar (put10) on every metric. Drop.
• 70_30 conservative collar (put15) dominated by 70_30 tight collar (put10). Drop unless put-pricing risk specifically pushes back.
• 50_50 collar (put10) dominated by 50_50 collar (put05) on every metric. Drop.
• 0_100 dominated by 50_50 collar (put05). Drop.
• 70_30 collar (put10) not dominated. Best Sharpe (1.46), best DD (−47%), but lowest NAV among survivors ($73M).
• 50_50 collar (put05) not dominated. Best NAV ($212M), highest trough ($18.2M), Sharpe 1.43. DD −55% (8pp worse than 70_30 tight).

The two survivors form the actual Pareto frontier — neither dominates the other. The choice between them is a personal trade-off between NAV and DD%.

Section 2.5 · Wipeout constraint

The hard constraint added late: no plausible path to wipeout under any adverse regime, including ones not seen in 2019–2026. This section asks two questions: (a) how real is the deep-impairment risk for the X3 sleeve? (b) what survives the constraint?

Calibration note (2026-05): X3 model recalibrated against the real Avanza BULL TESLA X3 cert chain. Default annual drag now 0% (real cert matches synth-no-drag within 2pp over 6yr). Removed single-day knockout — real cert never zeroed even at −97% drawdown (AVA 2). Deep impairment is asymptotic, not literal zero — but the sleeve has gone −90% (AVA 1, 2020-2026) and −97% (AVA 2, before delisting) in real life.

Mechanism & stylized stress test

The MA10-gated X3 sleeve loses value via cumulative chop-decay across many MA10 whipsaws. (Single-day knockout at TSLA <−33% has never been observed — worst was −21% in Sep 2020 — and the real cert doesn't zero out anyway; it asymptotes.)

Bar heights = % of 1000 sims (post-recalibration X3 model: drag 0%, epsilon-floor instead of knockout). Even at TSLA's historical mu=50%/vol=55%, 10% of forward 24-month windows leave the X3 sleeve at <10% of starting value. Under chop, the rate climbs to 31–61%.

Why this happens (mechanism):

• X3 daily-reset decay. A 3× daily-rebalanced product loses value to volatility: per-cycle decay ≈ σ²·(L²−L)/2 = 3·σ². At TSLA's 55% vol, that's ~90% annualized drag if continuously deployed in pure flat — compounding against you.
• MA10 gate cuts engagement to ~50%. Below MA10, sleeve sits in cash — that halves the drag exposure. But it also halves the win exposure.
• Whipsaw cost. Each MA10 cross-up + cross-down pair costs 0.5% in transaction fees. Pure flat with 55% vol generates ~12 round-trips per year → ~6%/yr cost just in fees.
• Compounding. Several sub-1% losses per cycle, sustained over 24 months, compound to 50–90% loss — the dominant tail mechanism. The real Morgan Stanley cert never zeros, but it does asymptote (AVA 1 hit −95%, AVA 2 hit −97% before delisting).

Plausibility — could it actually happen?

• Has not happened in TSLA's 7-year window. The closest historical analog (2021 mid-year chop, TSLA roughly flat 8 months) cost the X3 sleeve −18.5%. Extrapolated to 24 months ≈ −47% loss. Still far from wipeout, but materially worse than spot.
• Pre-2019 TSLA had multi-year flat regimes. 2014–2016 traded sideways for 2+ years with high vol. If that regime recurs, the X3 sleeve enters the wipeout zone.
• Other vol-heavy stocks have done it. Leveraged daily-reset ETFs on volatile single names (e.g. UVXY, JNUG) have had multiple >90% chop-decay episodes since 2010. Mechanism is identical.
• Yusuf's "live with this for years" framing means betting on TSLA not entering an unfamiliar regime in the next 10–20 years. Credible base rate: 10–25% per decade.

Safe-half discovery — stock-only collar (cc05 + put05)

Tested as the wipeout-immune anchor for two-level structures. No X3 sleeve = no chop-to-deep-loss path. Tight collar (cc05 + put05) keeps drawdown low and is already nearly cost-neutral (CC premium ≈ put cost).

Variant spread (robustness): NAV $10M–$32M across 6 timing variants. MaxDD remarkably stable at −20.7 to −22.5% (every variant). Sharpe 1.88–2.50 (every variant). Worst variant ($10M, last_close_open) is still 20× starting capital with the same DD profile.

Why it works:

• CC premium ($24.6M total) almost exactly funds put cost ($24.4M) — naturally cost-neutral, no Phase-5 backtester change needed.
• Tight cc05 cap means many cycles get assigned at +5% (44% assignment rate) — locks in steady gains.
• Tight put05 hard-floors the downside per cycle at −5% from spot. Compounded over the 2022 bear: drawdown bounded at −21% (vs −74% for stockonly, −55% for Pick A).
• $3K/cycle withdrawal funded entirely from CC premium — no asset sales.

Two-level structures — combine safe + leveraged

For users who want more NAV than stock_only_collar's $25M, allocate a fraction to a leveraged-collar finalist. The safe half remains wipeout-immune; the leveraged fraction is the only wipeout-able exposure. Tested four rebalance modes per split — asymmetric annual lock-in wins on every split.

Green = wipeout-immune anchor. Red = standalone leveraged collars. X marks = drift, filled circles = annual lock-in. Blue = Pick A risk half, purple = Pick B. Lock-in points sit ~10pp left of drift points at the same NAV — same return, lower drawdown. Pareto frontier under wipeout constraint runs through the lock-in points.

Pareto under wipeout constraint: the frontier becomes a continuum. You pick the X3 exposure you can stomach; NAV scales accordingly. Three natural picks emerge:

• Conservative anchor: 100% stock_only_collar — $25M, −21% DD, 0% wipeout exposure. Best Sharpe (2.36).
• Balanced two-level: 50% safe + 50% Pick A with annual lock-in — $118M, −40% DD, 25% wipeout exposure. NAV is 4.7× the conservative anchor; DD almost 2× as deep.
• Aggressive two-level: 30% safe + 70% Pick A with annual lock-in — $157M, −44% DD, 35% wipeout exposure. Trades more wipeout risk for ~33% more NAV than balanced, with similar DD.

Section 3 · Recommendation (under wipeout constraint)

Three picks survive once the wipeout constraint is applied. They sit on a continuous frontier between max-NAV and zero-wipeout-exposure. Pick by your appetite for the X3 chop-to-deep-loss tail.

How to pick between 1, 2, 3

My recommendation if forced to one: Pick 1 (stock_only_collar). Reasoning:

• Your stated hard constraint ranks wipeout-avoidance above NAV. Pick 1 is the only standalone candidate that satisfies it.
• Sharpe 2.36 means each unit of risk taken is more efficiently rewarded than any leveraged candidate. The leveraged options buy NAV at disproportionate risk cost.
• $25M is 50× starting capital — enough to fund $3K/mo withdrawal indefinitely with massive surplus. Marginal value of going from $25M to $118M is much less than from $0.5M to $25M.
• If after a few years your wealth + risk tolerance support more aggression, you can add a leveraged-collar sleeve later (effectively transitioning to Pick 2 or 3). The reverse is harder — once you've taken X3 wipeout exposure, you can't undo a wipeout that already happened.

Strong second: Pick 2 (50/50 two-level). If you find $25M too modest given the wealth this period generated for unleveraged TSLA bulls, the 50/50 two-level limits wipeout exposure to 25% while reaching $118M with MaxDD only −40%. Asymmetric annual lock-in keeps the structural protection intact (never moves capital from safe to risk).

Why all three beat your current strategy

• vs current 70_30_trim: trim is dominated (Section 1.2). Even Pick 1 marginally beats it on NAV (~$25M vs ~$27M) with vastly better DD (−21% vs −77%) and higher Sharpe (2.36 vs ~1.0). Picks 2 and 3 add 4–6× more NAV.
• vs Section 2 finalists (50_50 / 70_30 collars): those have higher historical NAV but unbounded wipeout exposure. Picks 2 and 3 retain most of the NAV while structurally bounding wipeout-able fraction.
• vs 0_100: dominated on every metric by Pick 2 (less NAV but vastly less DD and zero wipeout in the safe half). Hard-disqualified by the constraint anyway.

Regime override table

Subjective overrides Yusuf can apply when conditions warrant. Default = Pick 1 unless you choose 2 or 3 in advance.

Caveats and open questions

• Wipeout-tail estimates depend on TSLA's future regime distribution. Stress test assumes 24-month windows of fixed (mu, vol). Real TSLA regimes shift; a 6-month chop followed by a recovery doesn't wipe like a 24-month sustained chop. The 12% historical-mu wipeout estimate is a useful upper bound for sustained-chop concerns, not an unconditional probability.
• Two-level "drift" assumes no manual intervention. If you re-balance manually after a big move, you're not running pure drift — you may be inadvertently activating the lock-in mode that test showed hurts NAV. Stick to the structural design or change the design.
• Synthetic pricing — mostly N/A. Real-options chain covers Jan 2019 → Apr 2025 (~86% of cycles). Synthetic kicks in only for the post-Apr-2025 tail and uses the calibrated σ model (median +5% bias vs market, vs legacy +19%). Pure-synthetic gap on the collar would be ~24-34% NAV overstatement on legacy 1.12 — but this only shows up in the ~14% synthetic tail of the data, so end-state NAV impact is <1%.
• Put strikes not exhaustively tested. put05, put10, put15, put20 covered; put25+ untested. Tighter than put05 not tested but real-options chain enforces realistic pricing for <5% strikes.
• Initial capital sensitivity. Backtest starts at $500K. Options contracts have 100-share granularity — coverage % may not match exactly at very small bases. Pick 1's stock_only_collar is the most contract-granularity-friendly (single sleeve, no cross-sleeve allocation).
• Tax modeling. Schablonskatt only. Real-world CC assignment / put expiry create tax events not modeled. Pick 1 has higher CC assignment count (44% per cycle × 88 cycles = ~38 assignments) — more taxable events than the 70_30 chassis.
• Two-level sleeve management is operationally heavier. Pick 2/3 require two parallel cycles per month (one for each half). Pick 1 is a single cycle. Real-world execution friction not modeled.

Practical notes for switching

• From 70_30_trim to Pick 1 (stock_only_collar): liquidate the X3 sleeve over a few cycles (don't dump on a bad day); convert to 100% TSLA stock; start writing 5% OTM CCs against the full stock sleeve and buying 5% OTM puts using CC premium for funding.
• From Pick 1 to Pick 2 (50/50 two-level): liquidate half the stock to fund the leveraged half (50/50 stock/X3 sleeve with same cc05/put05 collar). Run two parallel monthly cycles.
• Annual lock-in for Pick 2/3: at year-end, check each half's value. If the leveraged half exceeds its target fraction (50% or 70%), move the excess back to the safe half. Never move from safe to leveraged — that's the asymmetry that preserves wipeout protection.
• Order of operations per cycle: write CC first to collect premium, then buy put. Same on each half if running Pick 2/3.
• If transitioning from a held trim-mode portfolio, the X3 sleeve is likely under target. For Pick 2/3, build the leveraged half over the next few cycles rather than all at once — avoids buying X3 at a possibly bad moment.
• Cert choice (Pick 2/3): hold AVA 1 (orderbookId 1041282), not AVA 3. AVA 1 outperformed AVA 3 by ~5.4%/yr over their 3.3-year overlap despite a wider spread (~0.31% vs 0.03%) — drag advantage exceeds spread cost.