About DawnScan

What is DawnScan?

DawnScan is a US stock monitoring service that automatically scans 6,000+ stocks listed on NASDAQ and NYSE after market close every day to capture technical preparation signals that appear before a surge (pre-breakout early signals).

The goal isn't to flag stocks that have already run up a lot — it's to surface stocks that look like they're right before a breakout, based on things like volatility contraction, accumulation, and relative-strength outperformance. Scan results come with AI-generated reasoning in Korean, and the hit rate is disclosed transparently — both hits and misses.

Who Runs This, and Why

This is built and run solo by an individual developer who goes by Dawn. My day job is IT engineering — I'm not in finance and I'm not an analyst. Since I'm not qualified to recommend stocks, I don't — instead, I measure the numbers and publish them.

The reason I built this is simple. There's no shortage of stock information out there, but almost nowhere actually checks "okay, but was it actually right" and publishes the answer. For any signal to credibly count as a "pre-breakout early signal," it has to be compared against the base rate — how often things play out the same way when that signal is absent. I run that comparison automatically every day, and publishing both the hits and the misses is the entire point of this site.

What Makes This Credible

• The denominator isn't hidden. Instead of a hand-picked set of winners, I log the entire universe of stocks that pass the liquidity filter every day. The current matured sample is 5,000+ observations, with an overall base rate of about 31% (reaching +15% within 20 trading days).
• Delisted stocks stay in the sample. Dropping stocks that disappeared and keeping only the survivors creates survivorship bias that makes the numbers look better than they are.
• Unfavorable results don't get erased. In August 2026, 2 signals that had been marked as statistically significant lost that effect once normalized by volatility (ATR) — meaning they were mostly just showing up on stocks that were volatile to begin with, not adding real signal on top of that. So I pulled that label myself and wrote up exactly what happened on the base rate proof page.
• What's unknown is labeled unknown. Signals without enough sample size are grayed out, and unverified numbers are never dressed up with phrasing like "N times the gain."

What This Site Doesn't Do

• Recommend stocks or provide trading signals — never
• Sell paid picks or subscription calls — never
• Promise returns, or use phrasing like "a high-probability setup" — never

DawnScan is an informational reference tool. It does not provide investment advice, asset management, or financial brokerage services. All decisions — and their outcomes — are the user's own responsibility.

Methodology Summary

The scanner scores stocks using a combination of the following technical signals.

• Bollinger Squeeze — detects volatility contraction (energy compression)
• OBV Accumulation — supply/demand leading a price move during consolidation
• Relative Strength (RS) — price outperformance versus the S&P 500
• VCP Base — convergence of price range below resistance
• RSI Pullback — a cooling-off zone setting up for a bounce without being overheated
• ADX, Volume, Gap Up — detects trend strength and supply/demand anomalies

The full algorithm is published on the methodology page.

Data Sources

• Price and volume (OHLCV) — yfinance (Yahoo Finance's unofficial API, up to 20 minutes delayed)
• Dividend data — yfinance
• AI-generated reasoning — an LLM (large language model) via litellm
• Stock universe — self-collected from stocks listed on NASDAQ and NYSE

yfinance is an unofficial data source for Yahoo Finance, so some stocks' data may be delayed or missing. It is not real-time data, and the accuracy of any figure is not guaranteed.

Editorial and Verification Principles

DawnScan is run on one principle: verify claims with data before publishing them. In practice, that means:

• Results aren't cherry-picked — every stock the scanner has ever surfaced has its subsequent performance published on the hit rate page, hits and misses alike, with nothing excluded.
• Everything is compared against a baseline — a signal's hit rate on its own tells you very little, so it's always benchmarked against the base rate of the entire universe, independent of the signal, and published on the base rate proof page. This applies a Wilson 95% confidence interval and a Benjamini-Hochberg multiple-comparison correction, and signals without enough sample size are labeled "insufficient sample" rather than stated as fact.
• Survivorship bias is removed — excluding delisted or halted stocks from the sample would inflate performance by leaving only the survivors, so they're included in the count.
• Unverified data is labeled as unverified — measurement-stage metrics whose base rate hasn't been confirmed yet (e.g., the crypto scanner) are explicitly marked as "measurement stage" on the page.
• Numbers refresh daily — scan results, hit rates, and base rates are all updated automatically every day, with the data source and delay (up to 20 minutes) disclosed as noted above.

If you spot an error, please let me know via the contact page. Errors in figures or explanations are corrected once verified.

SNS / Channels

Each channel posts different content.

• Instagram — @dawnbrief · The main channel. Scan-result cards and base-rate findings go up as short-form video (Reels).
• YouTube — @dawnbriefkr · The same videos are cross-posted as Shorts.
• Threads — @dawnbrief · Short text notes on the numbers and that day's observations.
• Telegram — @dawnbrief · Get the US market close briefing and data alerts.
• X — @dawnbriefdaily · An account exists, but regular posting mainly happens on the channels above.

No channel ever posts stock recommendations or trading signals. Only measured numbers and their limitations are shared.