Why a kurtosis model caught only part of HOOD's big run

HOOD was one of the period's largest movers, and the point of interest is what a choosy model does with a run like that: it banked a respectable share and walked past most of the rest. What follows runs the model's fixed rules forward across HOOD's saved history as a study of those rules, not a forecast or a recommendation; the live sample is still too small to grade.

What the model hunts for is clustering at HOOD's extremes. It asks whether the most violent moves, the thick edges of the recent range, have started bunching up, and it engages only when that bunching reaches an unusual peak. A peak that leans up buys and one that leans down sells, though on HOOD it leaned to the buy side thirty-four times out of forty. The rule cleared winners on 55% of its 40 trades and stacked up to +238.16% at a flat stake.

How the model is built, end to end.

The reading the model lives by is how clustered HOOD's largest moves have become. When ordinary days dominate, the edges of the recent move-size range stay thin and the reading stays low; when big up-or-down sessions start landing close together, those edges thicken and the reading climbs. It is scaled against how this stock normally behaves, so a level only counts as extreme relative to HOOD's own past. A quiet reading keeps the model out; only a pronounced thickening at the edges lets a trade form.

An extreme reading is a permission slip, not an order. The model still needs a direction read to agree before it acts, which keeps it from chasing every violent session. When both align it places a trade with a stop and a target fixed up front. After that the trade ends in one of three places, and on HOOD the stop did most of the work: the protective stop closed far more positions than either the target or the settling reading did.

The illustration above walks the model across HOOD in a single line: the edge-clustering reading, the gate it must pass, and the position it produces. None of the parts is fancy; the whole point is patience, waiting for the extremes to bunch up and the direction read to line up before any capital moves. The model's lineage belongs in the picture too: it began as one option among many in an automated search and survived only by clearing backtest and walk-forward checks, the road from first idea to working deployment behind every number on this page.

At a glance

These figures come from a backtest of the model on HOOD, scored against fixed acceptance gates, not from a live track record.

Walk-forward verification

A trade walked through

The walked example is a long held about three weeks. The edge-clustering reading pushed to an upward peak near 113 dollars, the direction read agreed, and the model bought; it sat through the chop and closed at its profit target near 142 dollars for +26.20%. It is the model at its best, patient entry on a clustered extreme, exit at the target once the move played out.

What the full trade record shows

Across its 40 HOOD trades the model won 22 and lost 18. The exits leaned on the stop: 17 were stopped out, 13 closed when the clustering reading settled back to ordinary, and 10 reached the profit target.

A stop-heavy split like this is the signature of a model that buys violent extremes and is often early: many positions get knocked out before the move it expected arrives, while the winners that survive are carried either to the target or out on the settling reading.

The biggest winner was a long that ran to its target for +35.51% over about eleven days off a low base in late 2024. The patient winners closed on the settling reading rather than a hard target, taking what the clustered extreme gave before it faded.

The worst trade was a long stopped out for -15.3% over two weeks in April 2026; that single loss is also the equity curve's deepest drawdown, and it is the reason the drawdown gate is the one that bites.

No single trade carries the record, and with HOOD up more than fourfold over the window, the lesson of the book is everything the model left on the table by waiting for the extremes to cluster before it would act.

How does this compare to just holding HOOD

Over the same window the model was tested on, simply buying HOOD and holding it would have done far better. Comparing the two shows whether the rule earned its keep or merely tagged along on a strong tape. Here the honest answer is that it tagged along, capturing a useful but partial slice, and the tiles below put numbers on the gap.

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 it was not built on. 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 says nothing about whether HOOD will keep clustering its extremes the way it did across the test window.

In live trading the model has been quiet so far, with too few signals to line up against the backtest. Until more live trades accumulate, the backtest is the only evidence there is, and it should be read as exactly that, a study of how the rules behaved on saved history.

When this approach fails

The model's failures come from the same place as its trades, the extremes. Its sharpest one is being early: the edges thicken, the model buys, and HOOD keeps falling, which is exactly how the deepest trade lost 15.3% on a long stopped out in April 2026. Because it is long most of the time, it has little defense against a sustained slide, and a stock this volatile can keep moving against a position longer than the reading expects. Its quieter cost is selectivity itself, every stretch it judged too ordinary to trade was upside in a fourfold run it never touched.

Three things are worth watching if this ever trades at size. The first is drawdown: the worst stretch in the backtest was about 15.3%, and the high stop-out rate means losses can arrive in clusters. The second is the long-heavy exposure, because the model has little way to profit from or hedge a sustained decline. The third is the distance between live and backtested behavior, the first sign the edge-clustering reading is no longer marking the same moves.

Risk and honest limits

On this run the model's automated checks logged a caution rather than a clean pass. It is one more reason to read everything here as backtest evidence about the rules, not a verdict on the stock.

Lifecycle