Research

Measured results, including the ones that went against us. Updated when the underlying analysis changes — not daily. For today's catalysts see the news page.

Do catalyst types predict excess return?

Every detected catalyst is graded against what the stock actually did over the following 60 trading days, measured as excess return versus SPY so a rising market does not flatter the result.

⚠️ Read the symbol-demeaned column. The raw column is dominated by which symbols happen to carry a catalyst type — per-symbol excess spans roughly +112% to −51%, so a type concentrated in strong names looks predictive whatever the catalyst was. Subtracting each symbol's own mean removes that.

Catalyst type	Raw	Symbol-demeaned	n (events)
earnings	+17.76%	+2.10%	138
partnership	+9.24%	+0.18%	204
regulatory	-16.20%	+0.10%	55
institutional	+14.73%	+0.05%	108
contract	-5.53%	-5.11%	86
product	+5.26%	-5.24%	48
macro	-0.99%	-7.70%	8
insider	-6.82%	-16.82%	11

Verdict: no catalyst type is tradeable on this evidence. The largest demeaned effect is earnings at +2.10% (n=138, 658 events total) — inside noise for a sample this concentrated. Most of the apparent signal in the raw column is symbol beta.

Strategies we tested and rejected

A published negative result is worth more than an unpublished positive one. These are our own ideas, tested against a pre-registered bar and reported as they came out.

Triple-EMA trend score (the signal behind our own trend charts) — FALSIFIED

We tested the signal that generates the trend charts on this site: does trading it beat simply holding the same instrument? Two directions were pre-registered before running — trend-following and contrarian — against a gate of beating buy-and-hold in 6 of 8 standard market periods.

Evidence. SPY: 2/8 trend-following, 0/8 contrarian. Across all 58 published ETFs, 116 arms were tested and exactly 1 cleared the gate — while chance alone predicts about 16.8. An earlier count of 31 'winners' collapsed once we required a period to be profitable and not merely better than a decaying inverse ETF: one arm scored 8/8 while making money in 0 of 8 periods.

Takeaway. The charts remain a useful picture of market state. Trading the score is not supported by the evidence, and we say so rather than quietly shelving it.

How we test

Pre-registration. The bar is fixed before the test runs, so a result cannot be reverse-engineered into a success.
A control that binds. A strategy is compared against buying and holding the same instrument over the same window — not against cash, which any long-biased strategy beats in a rising market.
Profitability, not just relative performance. Beating a losing benchmark means losing less money, which is not an edge.
Multiple testing counted. Test enough variants and some will pass by chance; we report how many were tried alongside how many survived.

BayTradar publishes market research only. No positions, portfolio holdings, or account data are disclosed. Nothing here is financial advice.