The Strategy: Momentum + AND-Gate Defense
A two-part TSP system. A momentum ensemble picks the strongest equity fund; a dual-signal AND gate — VIX backwardation AND SPY below its 200-day SMA — controls defensive positioning. Only when both signals agree does it move to G Fund.
Current Signal
BullAND Gate Verdict
BULL
0.8337
threshold 0.95
+10.3%
SPY $745.64 · SMA $676.25
C Fund ICS
+14.9
F Neutral
S Fund ICS
+14.9
F Neutral
I Fund ICS
+14.9
F Neutral
Updated 2026-05-22 22:13
How the Strategy Works
The system has two independent parts that work together:
- Momentum Ensemble picks which TSP equity fund to hold — the one with the strongest trend across multiple timeframes. Never holds the G or F Fund by choice.
- AND-Gate Defense decides whether to hold it. The gate goes defensive only when two independent stress signals agree:
- VIX/VIX3M ≥ 0.95 (volatility term-structure backwardation)
- SPY < SMA(200) (S&P 500 below its 200-day average)
The pattern the strategy is trying to capture: slow bear markets show up as SMA breakdowns; fast crisis shocks show up as VIX backwardation; genuine multi-month equity declines typically trigger both together. Requiring both signals filters out the false alarms each produces alone.
Part 1: Momentum Rotation
Ensemble Vote
Seven momentum calculations vote on which fund (C, S, or I) to hold, weighted toward a 63-day lookback as the core timeframe with shorter and longer windows for robustness. The fund that wins the majority vote gets 100% of the allocation. When the ensemble disagrees, the model defaults to C Fund (S&P 500) — historically, staying in large-cap U.S. equities beats sitting in cash during uncertain periods.
Equity-Only Rotation
Only C (S&P 500), S (small/mid-cap), and I (international) compete. F Fund (bonds) is excluded — bonds dilute the equity momentum signal. G Fund is never selected by momentum. The gate handles defense; momentum handles offense.
- 2017: 49% I Fund — international led, the model followed
- 2020: 58% S Fund — small caps led the recovery
- 2025-2026: Heavy I Fund as international momentum builds
Friday Rebalancing
End-of-week rebalancing captures the full week's price action. The model checks every Friday but only acts when the leader changes or when the gate flips. Average: under 1 transfer/month, well within the 2-IFT limit. Submit before noon ET Friday for same-day execution.
Part 2: AND-Gate Defense
The gate uses two independent stress signals. Each alone has false-alarm problems. Together, they filter each other.
Signal 1: VIX/VIX3M Ratio (volatility term-structure)
The ratio of 1-month to 3-month implied volatility. When it rises above 0.95, the near-term volatility premium is elevated relative to the forward premium (backwardation) — the market is pricing in imminent stress more than long-term stress.
Strength: catches sudden shocks (2020 COVID, April 2026 tariff shock). Weakness: false alarms on transient VIX spikes that don't reflect real trend breakdown (e.g., single-day earnings scares, brief geopolitical jitters).
Signal 2: SPY vs 200-day SMA (trend regime)
A classic trend-following signal: is the S&P 500 above or below its 200-day moving average? Below = trend is broken. This is a slow signal by design.
Strength: catches slow grinds (2022 bear market, 2008 GFC). Weakness: false alarms on whipsaws near the crossover (2011 Aug/Sep, 2015/2016 chop).
AND Logic: Both Must Fire
The gate only goes defensive when both signals agree. This eliminates each signal's worst failure mode:
- VIX spike without SMA break → stay invested. The trend is still intact; vol was a blip.
- SMA break without VIX stress → stay invested. Likely a whipsaw, not a real crisis.
- Both fire together → real bear market. Move to G Fund (free transfer).
| Signal State | VIX ≥ 0.95 | SPY < SMA(200) | Action | Historical frequency |
|---|---|---|---|---|
| Bull | No | No | 100% momentum fund | ~80% |
| VIX-Watch | Yes | No | Stay invested | ~10% |
| SMA-Watch | No | Yes | Stay invested | ~6% |
| AND-Bear | Yes | Yes | 100% G Fund (free) | ~9% |
Over 15 years of backtest, the AND gate is defensive about 9% of weeks vs 17% for VIX-only. Half the defensive activity, and almost all of it is concentrated in genuine bear regimes.
Backtest: 2010–2026
AND gate is the primary strategy (highlighted). VIX-only, SMA-only, pure momentum, and C Fund buy-and-hold are shown side-by-side so you can see what each signal contributes.
| Metric | C Fund (B&H) | Pure Momentum | + VIX only | + SMA only | + AND Gate |
|---|---|---|---|---|---|
| Total Return | 860.2% | 1247.7% | 1228.2% | 853.0% | 1313.5% |
| Annualized Return | 15.3% | 17.8% | 17.7% | 15.3% | 18.2% |
| Sharpe Ratio | 0.913 | 0.986 | 1.291 | 1.106 | 1.192 |
| Sortino Ratio | 1.123 | 1.263 | 1.604 | 1.348 | 1.526 |
| Calmar Ratio | 0.454 | 0.523 | 0.678 | 0.611 | 0.567 |
| Max Drawdown | -33.7% | -34.0% | -26.1% | -25.0% | -32.0% |
Read the table carefully: VIX-only has the highest Sharpe (best ride), but AND gate has the highest total return (best wealth). The tradeoff is a modest Sharpe haircut in exchange for ~85 percentage points more cumulative return over 15+ years. For a long horizon, total return wins that tradeoff.
Important: Backtests are not predictions. They show how a strategy would have performed using historical data. Markets change. Use this data to understand the approach, not as a guarantee.
The 2008 GFC Stress Test
Extended backtest (2004–2026, 22.5 years) captures the 2008 Global Financial Crisis. Peak-to-trough from Oct 2007 to March 2009:
| Strategy | GFC Period Return | Peak-to-Trough Drawdown |
|---|---|---|
| C Fund buy-and-hold | -45.5% | -55.2% |
| Pure momentum (no gate) | -54.0% | -62.8% |
| VIX-only gate | +3.4% | -7.3% |
| AND gate | +5.8% | -7.3% |
Pure momentum (no defensive gate) lost more than buy-and-hold during the GFC — rotating between C, S, and I is useless when all equity crashes together. Any plan with a 15+ year horizon must assume another GFC-scale event will happen. The gate is not optional.
2026 YTD — What the AND Gate Actually Did
Live tracker started January 2026 at $100,000. Q1 saw tariff-driven volatility that tested the gate's discipline:
| Period | Signal | What happened | AND gate response |
|---|---|---|---|
| Feb 5–6 | VIX-Watch | VIX spiked to 21.8 on tariff anxiety, but SPY held above SMA(200) | Stayed in I Fund — avoided VIX-only whipsaw |
| March 2–19 | Mostly Bull / brief VIX-Watch | VIX flickered near 0.95 threshold | Stayed invested — no trend break |
| March 27–April 8 | AND-Bear | Full tariff shock. VIX hit 1.06 (deep backwardation) AND SPY broke below SMA(200) | Moved to G Fund — avoided 15–20% equity losses |
| April 9–present | Bull | Tariff pause; both signals clear again | Re-entered I Fund |
The AND gate's discipline shows clearly: the Feb and early-March VIX flickers would have dragged a VIX-only gate into G Fund and missed the intervening rallies. The AND gate required the SMA signal to confirm, and it didn't until the real tariff shock in late March. That's the design working as intended.
Growth of $100,000 (2010-2026)
AND gate (solid purple) vs VIX-only (dotted) vs C Fund (dashed). Gold area = AND gate's % in G Fund.
Growth of $100,000
Strategy vs Buy & Hold vs C Fund Only (2010-2026)
Fund Allocation by Year
The strategy actively rotates — it's not a disguised C Fund buy-and-hold
Alpha by Market Environment
Excess return vs buy-and-hold in every sub-period — positive in all 11
Monte Carlo: Strategy vs 10,000 Random Simulations
Sharpe ratio distribution — the strategy (0.956) beats 100% of random fund picks
Validation: Is This Overfitting?
Walk-Forward Optimization (12 folds, 3yr train / 1yr test)
Each fold optimizes both gate thresholds (VIX 0.85–1.05 × SMA period 150–250) on a 3-year training window, then tests the winning combination on the following year out-of-sample. Parameters were never touched during test periods.
OOS Avg Sharpe
1.316
OOS Avg MaxDD
-7.8%
OOS Avg Return/yr
8.0%
Beat C (raw ret)
25%
| OOS Period | VIX thr | SMA | Return | Sharpe | MaxDD | vs C |
|---|---|---|---|---|---|---|
| 2013-07-08 → 2014-07-07 | 1.05 | 150 | 4.8% | 0.815 | -6.6% | -18.0% |
| 2014-07-08 → 2015-07-07 | 0.90 | 180 | 6.2% | 0.973 | -4.7% | -1.9% |
| 2015-07-08 → 2016-07-06 | 1.05 | 150 | 7.0% | 0.857 | -8.0% | +2.3% |
| 2016-07-07 → 2017-07-06 | 0.95 | 220 | 9.0% | 1.980 | -2.8% | -8.1% |
| 2017-07-07 → 2018-07-06 | 0.95 | 200 | 5.7% | 0.772 | -10.1% | -10.2% |
| 2018-07-09 → 2019-07-09 | 0.95 | 200 | 13.5% | 2.257 | -5.4% | +4.5% |
| 2019-07-10 → 2020-07-08 | 1.00 | 150 | 12.6% | 1.224 | -12.4% | +4.6% |
| 2020-07-09 → 2021-07-08 | 1.05 | 150 | 12.7% | 1.414 | -8.5% | -26.5% |
| 2021-07-09 → 2022-07-08 | 1.05 | 150 | -17.1% | -1.374 | -22.6% | -7.6% |
| 2022-07-11 → 2023-07-11 | 0.90 | 200 | 14.0% | 2.378 | -5.0% | -3.0% |
| 2023-07-12 → 2024-07-11 | 0.90 | 150 | 15.2% | 2.293 | -5.3% | -11.3% |
| 2024-07-12 → 2025-07-15 | 0.90 | 150 | 12.0% | 2.206 | -1.9% | -0.5% |
OOS Sharpe averages 1.316 across 12 independent 1-year windows. The strategy beats C Fund on raw returns in 25% of windows — this is honest. The remaining windows are mostly bull-market runs where 100% C Fund is unbeatable on raw return; a defensive strategy will always look bad there. The edge shows up in risk-adjusted performance and in bear windows: average OOS MaxDD of -7.8% vs C Fund's -33.7% over the full period.
Monte Carlo (1,000 simulations)
Bootstrap resampling of the strategy's 3,995 daily returns — tests whether the result depends on a lucky sequence of returns.
| Metric | Actual | p5 | p25 | p50 | p75 | p95 |
|---|---|---|---|---|---|---|
| Total Return | 1313.5% | 441% | 831% | 1259% | 2009% | 3703% |
| Sharpe Ratio | 1.192 | 0.777 | 1.014 | 1.176 | 1.367 | 1.636 |
| Max Drawdown | -32.0% | -34.2% | -27.8% | -23.4% | -19.9% | -16.6% |
Actual 1314% return sits near the p50 of 1,000 simulations. For a defensive strategy that's expected — the alpha lives in Sharpe and drawdown shape, not in terminal-return tail luck. Downside p5 scenario still returns 441% over 16 years.
Honest framing: The AND gate's edge over C Fund is not raw return — it's the combination of staying invested most of the time (momentum captures the bull-market beta) while sidestepping the deep, multi-month bear drawdowns that would otherwise compound forward. A defensive strategy that beats C Fund every year is fiction; an honest one loses in bull years and wins in bear years with enough margin that the long-run geometric mean comes out ahead.
Drawdown & Psychology
What a -33.7% Drawdown Feels Like
On a $500,000 TSP balance, a -33.7% drawdown (C Fund B&H's worst) means watching $168,000 disappear. That's over two years of GS-12 gross salary. Research from Dalbar and Vanguard shows that 35-50% of retail investors make panic-driven changes during drawdowns exceeding 30%. Kahneman's prospect theory puts losses at roughly 2x the emotional weight of equivalent gains.
What the AND Gate's -32.0% Drawdown Feels Like
AND gate's worst historical drawdown on a $500k balance is $160,000. Smaller doesn't mean painless, but it's the difference between a bad year and a career-defining crisis. Abandonment rates for drawdowns under 25% drop to 10–15%.
The Real Cost of Panic
If there's even a 25-30% chance you'd deviate during a severe drawdown, the expected value of gating exceeds pure momentum. One panic sell that misses half a recovery costs more than years of gate drag.
Frame it as insurance, not as drag. You don't cancel homeowner's insurance because your house didn't burn down. The AND gate is the premium. The payout comes during the -30% drawdown you don't experience.
TSP Transfer Rules
- Moves to G Fund are IFT-free — unlimited, don't count against your 2 IFTs/month. But the noon ET deadline still applies to all moves including G.
- 2 IFTs per month — the AND gate uses under 1/month on average.
- Noon ET deadline — all moves — submit before noon ET for same-day close prices. This applies to G Fund moves too.
- Contribution changes are free — redirect future contributions anytime without using an IFT. More here.
Who This Strategy Is For
- 15+ year horizon — the longer, the more drawdowns get absorbed by compounding.
- Systematic mindset — you prefer rules over gut feelings, and you can execute weekly.
- Federal employee — exploits TSP's unique structural advantages (free G moves, zero slippage at NAV, zero fees).
Read before you follow: This is a personal project tracking a quantitative strategy. It is not financial advice. Understand the drawdown risk, the mechanical nature of the allocation, and the limitations of backtesting before following any signals.