KOSPI6,687.21 +1.64%KOSDAQ813.50 +2.95%거래대금 21.92기준일 2026-09-04
Part 2 — Charts and Spikes · Myth Testing 6 · 2026-09-01 · ~4 min read← All research

Buying Before the Spike — Five Front-Running Signals, 820,000 Cases Tested

Machine-readable: Markdown · 한국어 원문

In the previous piece we confirmed that chasing after a spike is negative on every path. So what if you enter early, when the precursors of a spike (investor-flow accumulation, volume surges) appear? We tested this by not including spike occurrence in the conditions (blocking future information) and entering at every point where the precursor signal was observed.

Results — Five Front-Running Signals

Front-running signal (enter at observation, hold 20 days)SamplesExcess vs marketBeat-market rate
Foreign investors' 5-day accumulation ≥ 10% of volume273,780-0.22%p48.8%
Foreign investors' 5-day accumulation ≥ 20% (stronger signal)79,769-0.57%p47.1%
Institutions' 5-day accumulation ≥ 10%155,828+0.9%p55.5%
Volume surge of 2.5x198,966-1.12%p45.1%
Price +10% · volume 2x rising together beforehand82,813-3.73%p39.9%
Above condition + no identifiable buyer (institutions and foreign investors each under 1%)42,666-5.47%p35.3%

What These Numbers Say

Most of what looks like a precursor to a spike is not a precursor to a spike. Accumulation and volume surges are frequently found in hindsight among stocks that spiked, but the vast majority of stocks where the signal fires do not spike (the base rate problem).

The worst signal is a separate one. When price and volume rise together but no identifiable buyer is present, the result was worst at -5.47%p. That is exactly the pattern that looks like "quiet accumulation" if you only look at the chart.

Only the institutional accumulation signal (+0.9%p) was positive in direction, but given transaction costs it is hard to view as a real-world edge, so we judged it "negligible effect."

Look-Ahead Warning

Computing "entry on the day before the spike, among stocks that spiked" with the same data looks like +9.4%p. That is a figure that assumes you know in advance which stocks will spike — not strategy performance, but an upper-bound reference point. Many backtests claiming a surefire method for spike stocks contain future references of this form.

Data Sources

  • Korea Exchange daily quotes and investor-by-investor trading trends, full coverage. Investor flows per stock can be checked on each stock page

This article documents tests on historical data for informational purposes only. It is not investment advice or a recommendation to buy or sell any security. Past test results do not guarantee future returns.

Comments

Comments on methods, data and interpretation are welcome. Buy/sell recommendations for specific securities may be removed.