Quantitative Research
Test 3 — Is it just the bid-ask spread in penny stocks?
A round-trip buy-then-sell in an illiquid name can show a "return" that is really just bouncing between the bid and the ask, no informed trading involved. If that were the whole story, the effect should vanish once penny stocks are excluded.
The trade journal is barely profitable at a token $200 stake per trade. Before concluding the effect is too small to matter, three ways it could be a statistical illusion were tested directly against the raw data — a crash-clustering artifact, inflated significance from correlated same-day events, and pure bid-ask bounce in illiquid names. All three were built to make the finding disappear. None of them did.
Weakens, but survives. The return itself does shrink as price rises — consistent with some bid-ask noise in the cheapest names — but the t-statistic stays well above the ±2 significance threshold even restricted to stocks trading above $50, where the spread is negligible.
Verdict: the signal is real, but tiny. Three separate ways to explain it away all failed. What survives, however, is economically small: a fixed $200 stake per trade turns roughly 8,000 signals over 23 years into a low-single-digit-percent total return — nowhere close to compensating for a real trading operation's costs at that scale. The open question was never "is there an edge," it was "can normal position-sizing turn that edge into a return worth pursuing." That is what the diversified-portfolio page tests directly.