Why a selective momentum model trailed ASTS's runaway trend
It reads momentum across three time horizons and acts only when a supermajority agree, confirmed by a separate directional read. That confirmation kept it disciplined in normal conditions and cost it dearly in a vertical move: it returned +439.88% but a plain hold of ASTS returned +1799.80%. A 54% win rate with large winners is a normal momentum profile; the 20% drawdown is why it stays a backtest study, not a live forecast.
ASTS was one of the great runs of the period, almost nineteen-fold, and a momentum model that waited for confirmation captured a real slice of it but left far more on the table. This is a backtest of a rule-based ASTS strategy, not a live forecast, and the live record is still too short to judge it.
The model reads momentum over three time horizons and fires only when a supermajority of them point the same way, then takes the trade only if a separate directional check confirms it. Those gates are protection: they keep the model out of noise. In a stock that went almost straight up they were also a tax, because every bar spent waiting for confirmation was a bar of the move missed. The model won 25 of its 46 trades, and its per-trade returns added up to +439.88% on a fixed position size, carried by a few large winners.
How the model is built, end to end.
The model watches one monitored ASTS reading and measures its momentum over three look-back lengths at once: short, medium, and long. Each horizon votes on direction, up or down. The model acts only when a supermajority of the three agree, so one noisy horizon cannot move it; a split vote keeps it flat.
When the horizons agree on up and a separate directional read confirms it, the model opens a long; when they agree on down with the same confirmation, a short. From there the trade runs on fixed rails, a protective stop and a profit target. This is the opposite of fading extremes: the model is built to stay with a move, which is why its winners are large and its stops, when they come, are frequent. Roughly as many trades hit the stop as the target.
The illustration follows the model's lineage on ASTS end to end: from three momentum reads, through the agreement vote and the confirming gate, to the order that goes out. Each check is ordinary; the model's character is in requiring all of them to line up before it commits.
At a glance
| Gate | Actual | Threshold | Status | Threshold source |
|---|---|---|---|---|
| win rate | 54.35% | >= 70.00% | fail | canonical registry standard |
| max drawdown | 20.42% | <= 5.00% | fail | canonical registry standard |
| sample size | 46 | >= 30 | pass | canonical registry standard |
| total return | 439.88% | >= 100.00% | pass | canonical registry standard |
| expected return | 9.563% | >= 5.000% | pass | canonical registry standard |
| Metric | Value |
|---|---|
| Total return | 440% |
| Win rate | 54.3% |
| Max drawdown | 20.4% |
| Expected per trade | 9.56% |
| Trades | 46 |
These figures come from the model's backtest measured under backtest conditions and against the registry's fixed acceptance gates. The total-return figure is the sum of the model's per-trade returns at a fixed position size, not a compounded equity curve; the buy-and-hold benchmark, by contrast, is a compounded hold of the stock, so the two are different measures.
Walk-forward verification
| Metric | Value |
|---|---|
| Walk-forward match | 100% |
| Verified timestamps | 1,739 |
| Signal correlation | 1 |
A trade walked through
On a steady afternoon for ASTS, the model opened a long trade after several hours of patience. It waited through a shallow wobble in the wrong direction before the move it expected actually arrived. The position closed on a TP exit, with the broader market sitting in a mixed stretch, exactly the kind of regime the model was trained to read.
| Metric | Value |
|---|---|
| Direction | long |
| Entry price | 67.83 USD |
| Exit price | 89.87 USD |
| Hold time | 5.0 days |
| Return | +32.49% |
What the full trade record shows
Across its 46 ASTS trades the model won 25 and lost 21, all on the long side. The exits split three ways: 19 closed at the protective stop, 18 reached the profit target, and 9 ran to the holding limit. These are backtest results, not a forecast.
| Exit reason | Trades | Share |
|---|---|---|
| Stop-out | 19 | 41.30% |
| Take-profit | 18 | 39.13% |
| Time exit | 9 | 19.57% |
That near-even split between stops and targets is the mark of a momentum system in a volatile name: it is right often enough to pay, but it is stopped out almost as often as it hits its target.
The biggest winner rode the trend to the holding limit for +90.66%; a clean breakout added +31.00% in a day; a slower long closed at the time limit for +7.70%.
The worst trade was a false breakout that reversed into its stop for -18.40%.
The record leans heavily on a few large trend-following winners to offset a high stop-out rate, which is exactly why the drawdown and sample-size gates matter here.
How does this compare to just holding ASTS
Over the same window the model was tested on, simply buying ASTS and doing nothing was its own kind of strategy. Comparing the two tells you whether the model was earning its keep or just riding the tide. In this case the model underperformed a plain hold, and the tiles below show the size of the gap.
| Metric | Value |
|---|---|
| Model total return | +439.88% |
| Buy-and-hold | +1799.80% |
| Difference | -1359.92% |
How well does the model reproduce its tape?
Walk-forward verification checks whether the saved rule path reproduces the expected signal behavior on held-out timestamps that were not used to build it. It is a consistency and replay-integrity test, not proof the model will make money live. A clean reproduction means the deployed rules behave like the studied rules; it does not say ASTS will keep trending the way it did.
| Metric | Value |
|---|---|
| Match rate | 100.0% |
| Correlation | 1.000 |
| Alignment | Quiet |
On ASTS, the model is still quiet against its backtest distribution. Aligned means the model's live trades look statistically like the trades it took in training. Drifting means at least one statistic, hold time, hit rate, or exit type, has moved away from where the backtest sat, a sign the live record has started to diverge from the test. Quiet means there is not yet enough live evidence to call it either way. None of these labels prove the model good or bad; they describe how closely the live record is tracking the backtest so far.
When this approach fails
The model's failures are the failures of trend-following. The sharpest is a violent reversal after a speculative run: the gates confirm late, the model goes long near a top, and the move snaps back. Gap-downs on financing or news jump straight through the protective stop. False breakouts lure the vote into agreement just before the move fails. And in a crowded momentum name, the unwind can be faster than any exit rule. The flip side of large winners is that the losers cluster, which is how a 20% drawdown gets built.
| Metric | Value |
|---|---|
| Losing trades | 21 |
| Worst single-trade return | -18.40% |
| Worst in-trade drawdown | -20.42% |
Running this on ASTS live, a desk watches three things. The first is drawdown, not just the current trade's drawdown, but the rolling drawdown of the strategy over the past several weeks, and a hard rule that pauses the model when it crosses a pre-agreed line. The second is liquidity and financing: thin books and high borrow or financing costs turn small edges into nothing. The third is alignment with the backtest distribution, when live trades start looking nothing like the backtest sample, the model is telling the desk something has changed.
Risk and honest limits
The model is checked against a fixed verification pipeline at each release: contract integrity, signal reproduction, trade parity, and behavior. Its current lifecycle state is shown below.
Lifecycle
Where we are
This reflects the model's measured backtest. There is not yet enough live history to add a live-performance section; until there is, read every figure here as backtest evidence.
Sources
This article is based on Stonewell One research, including backtesting, walk-forward verification, deployment monitoring, and model-risk review.Trade-level entries, exits, and holding times come from Stonewell One's backtest of ASTS over the 2024 to 2026 replay window.The model is compared against simply owning ASTS over the same window.