“Buy before, sell the spikes, hold the drops”
Tested on all 580 reports across the watchlist. Buy at the last close before the numbers; if it gaps more than +5% sell into the first session; if it gaps down, keep holding. Compared against the same entries held regardless.
It costs 1.2% per report versus just holding, and it gets worse the longer you hold.
The rule is backwards on both halves. Big up-gaps carry on drifting up, so selling into the spike cuts short the trades that were working. Big down-gaps go on lagging, so holding them keeps exactly the trades that were not. It underperforms at every horizon tested, significantly so in each case, on 8 of 9 companies individually.
The catch that makes it feel right: it raises your hit rate, from 54% to 59% of reports profitable. More frequent small wins, less money — and it does nothing for the downside, since its worst case (−38.1%) is the same trade that always holding suffers (−38.1%). You cap the upside and keep all the risk.
Why — spikes run, drops linger
Measured from the close of the first session, so the gap itself is already in the price. The question is only what happens next, and it decides the whole strategy.
After a big up-gap the stock keeps up with, or beats, an ordinary hold at every horizon. After a big down-gap it lags one at every horizon — over 3 months, +6.9% against +9.3%. Down-gaps do not bounce back; that is the assumption the strategy rests on, and the data does not support it.
Four rules, same entries
Every rule buys at the same last-close-before-the-report on all 580 reports, so “vs holding” is a paired difference. A 95% range that stays below zero means the underperformance is more than noise.
| Rule | Exit | Average | Hit rate | Worst | Swing (sd) | vs holding |
|---|---|---|---|---|---|---|
| Hold regardless | 2 weeks | +2.1% | 55% | −35.7% | 12.2 | benchmark |
| Always sell straight away | 2 weeks | +1.1% | 54% | −26.4% | 8.4 | −0.9%[−1.6%, −0.3%] |
| Sell spikes, hold dropsyours | 2 weeks | +1.4% | 57% | −35.7% | 10.3 | −0.7%[−1.1%, −0.2%] |
| Sell drops, hold spikes | 2 weeks | +2.0% | 53% | −26.4% | 11.9 | 0.0%[−0.2%, +0.2%] |
| Hold regardless | 1 month | +2.6% | 54% | −38.1% | 14.6 | benchmark |
| Always sell straight away | 1 month | +1.1% | 54% | −26.4% | 8.4 | −1.5%[−2.4%, −0.6%] |
| Sell spikes, hold dropsyours | 1 month | +1.4% | 59% | −38.1% | 11.2 | −1.2%[−1.8%, −0.5%] |
| Sell drops, hold spikes | 1 month | +2.5% | 53% | −32.2% | 14.1 | 0.0%[−0.4%, +0.3%] |
| Hold regardless | 3 months | +9.0% | 64% | −58.1% | 23.9 | benchmark |
| Always sell straight away | 3 months | +1.1% | 54% | −26.4% | 8.4 | −7.9%[−9.7%, −6.2%] |
| Sell spikes, hold dropsyours | 3 months | +5.5% | 68% | −58.1% | 17.4 | −3.5%[−4.8%, −2.3%] |
| Sell drops, hold spikes | 3 months | +7.8% | 59% | −51.1% | 22.8 | −1.2%[−2.0%, −0.4%] |
Consistent across companies
The rule against always holding, over 1 month, on each company's own reports. A pooled result driven by one ticker would show up here.
What this test does not cover
- · No trading costs, spreads, slippage or tax. Selling into a spike is a taxable event that holding is not, which makes the real gap wider than shown.
- · These nine companies were chosen in hindsight for being famous successes, so every absolute return here is far too flattering. The paired “vs holding” column is the trustworthy part — it compares two rules on identical entries.
- · 580 reports is a decent sample pooled, but it is one market era. A regime where earnings gaps mean-revert would flip the conclusion.
- · Descriptive statistics on the past. Not advice, and not a forecast.