Suzaku is Shishin's momentum-breakout engine, one of four regime-specialist engines. It trades breakouts (a base, a pivot, an expansion) only in the market regimes where breakouts tend to follow through, and stands aside otherwise.

What does a breakout engine's win rate look like?

In Shishin's hypothetical five-year backtest, Suzaku won 43.5% of its 131 trades, fewer than half. It was still net positive because the average winner ran several times the size of the average loser: the edge is in the payoff shape, not the hit rate. Past performance does not predict future results.

How does Suzaku exit a trade?

It fills on the close, then holds until the position either closes below its 12-day moving average (trend bent) or hits a volatility-scaled stop (1.25x the 14-day ATR, floored at 2%). The median backtested holding period was about 11 trading days.

Inside Suzaku: Shishin's momentum-breakout engine, by the numbers

Q: When does Suzaku trade?

Only when a breadth-driven regime classifier opens the gate. In a constructive, broad tape it is handed the day; when breadth deteriorates it stands aside and the system holds cash or routes to another engine. The regime is the gate; the breakout pattern fires inside it.

This article describes one of Shishin’s research engines and the results it produced in a hypothetical five-year backtest. Backtest figures are not live trading results and do not predict future performance. It is educational, an explanation of how a systematic engine works, not personalised investment advice, and not a recommendation to buy or sell any security.

Suzaku (朱雀) is the breakout specialist among Shishin’s four engines. If how breakout setups work is the general theory, a quiet base, a pivot, an expansion, and a high failure rate, this is the specific implementation: when Suzaku is allowed to trade, what it looks for, how it gets out, and what its 131 backtested trades actually did. It is the clearest single window into how one of the engines really operates.

When Suzaku is allowed to trade

The first and most important thing about Suzaku is that it is usually switched off. Breakouts only follow through in certain market states, so a breadth-driven regime classifier sits in front of it as a gate. When the broad tape is constructive, Suzaku is handed the day; when breadth deteriorates, it stands aside and the system sits in cash or routes to a different engine entirely. The regime is the gate, the breakout pattern is what fires once the gate is open, the ordering that the evergreen pillar argues for, made concrete. It is also why a breakout engine can be patient: it is not trying to trade breakouts in tapes that punish them.

What Suzaku looks for

Inside an open gate, Suzaku scans the tradable universe and ranks candidates with the same composite score the rest of the system uses, weighing trend, momentum, position relative to moving averages, volatility, volume, and range structure into one number. The breakout-specific inputs are the ones the theory predicts: a long, contracting base; a clear on expanding range and volume; and a name that is freshly emerging rather than already extended, which is the job of the setup-state classifier. The score decides which of the eligible breakouts to take and how large to size them; the base, not the ticker, earns the rank.

How it enters and exits

Suzaku scores and fills on the close, not the intraday print, which quietly solves the two problems that sink most breakout traders. A stop-hunt that pokes above the pivot and reverses before the bell never produces an entry, because the close has to hold above the level to qualify. And a name that gapped and already ran scores as extended and ranks down rather than getting chased at the open.

Once in, the position is held until the first of two things happens: it closes below its 12-day moving average (the trend has bent), or it hits a volatility-scaled stop, 1.25× the 14-day ATR, floored at 2%, intraday (the breakout was rejected). That is the operational definition of “follow-through”: a trade followed through if it survived to the trailing-MA exit rather than the stop. In the backtest the split was almost even, 69 of the 131 exits were the trailing MA, 60 were the stop, two ran to the end of the test, and the median holding period was 11 trading days. Why a close-confirmed trail rather than a tighter intraday one is itself a tested choice, covered in volatility-aware stops.

What its trades actually look like

Here is every one of Suzaku’s 131 backtested trades, binned by outcome. This is the shape the theory predicts, made real:

Fig., Suzaku’s 131 breakout trades, by outcome (locked five-year backtest, hypothetical). Most trades are small losers cut by the volatility stop; the return lives in a thin right tail. Mean +10.5% sits far above the −4.0% median, the textbook signature of a positive-expectancy setup with a sub-50% hit rate. Past performance does not predict future results.

43.5% win rate. 57 winners, 74 losers. Fewer than half, a coin-flip would do better on hit-rate alone.
The median trade lost 4.0%, the typical breakout was a small loser, clustered where the volatility stop cut it (that tall −5% to −10% bar).
The mean trade still made +10.5%, because the average winner (+34.2%) ran more than four times the size of the average loser (−7.7%). The mean sitting far to the right of the median is the signature of a thin tail of large winners doing the work the sub-50% hit rate cannot.

That right skew is the definition of a breakout edge, not a flaw in it: you lose small and often, and win big and rarely. The natural follow-up , does an edge like this hold up when its luckiest names are stripped out?, is its own question, and we put the whole book through exactly that robustness test in leave out the winners. Every trade above is in the downloadable ledger in the five-year record.

What we tried to make it pick better, and didn’t keep

A reasonable instinct is that the engine’s edge should come from grading the pattern more finely. We tested that: adding a relative-strength percentile feature, scored within the universe, to sharpen which breakouts Suzaku preferred. It didn’t help, the feature turned out to be negatively related to forward returns over our sample, so we dropped it. The lesson, consistent with everything else in the system, is that the edge lives in whenthe engine is allowed to fire and how the failures are managed, far more than in ever-finer chart grading. More of those tested-and-dropped ideas are in the experiments that failed.

Why Suzaku is one of four, not the whole strategy

A breakout engine has regimes that suit it and regimes that punish it. Run alone, it would spend long stretches either forcing trades into hostile tapes or sitting idle. Pairing it with three specialists tuned to the other conditions, quality, defensive, and large-cap recovery, is what lets each one wait for its own weather without the overall book going quiet. Suzaku is the engine that pays off when momentum is working; it is explicitly not asked to be the engine for every market. That division of labour is the whole point of four engines for four regimes.

Sources & methodology

The figures here are the Suzaku slice of Shishin’s locked five-year backtest (hypothetical). How they are measured, survivorship-free universe, close-fill convention, the leave-out-the- winners robustness test, and the per-trade significance battery, is documented in the five-year record, leave out the winners, why backtests lie, and statistical significance. The setup theory and its external literature (Jegadeesh-Titman 1993; O’Neil / CAN SLIM; Minervini / VCP; Bulkowski) are collected in how breakout setups work.

Shishin Research

Inside Suzaku. The breakout engine, by the numbers.