AI Summary
This video tests four famous candlestick patterns (hammer, morning star, bullish engulfing, and three white soldiers) across SPY, QQQ, Bitcoin, and Ethereum using daily candles from 2018. The creator uses a strict methodology: wait for the pattern to complete, enter at the next open, hold for 5 bars, and exit, with 0.1% commission each way. The goal is to determine which patterns actually beat each market's normal 5-bar average return, rather than relying on cherry-picked examples.
Chapters
Four candlestick patterns tested across SPY, QQQ, Bitcoin, and Ethereum using daily candles starting in 2018. Entry at next open after pattern completion, hold 5 bars, exit. Commission 0.1% each way, no slippage simulation.
Hammer appeared 241 times with a 50.2% win rate but an average loss of 0.71%. Winning half the time doesn't help when losses are larger than wins.
Morning star had 195 appearances, a 44.1% win rate, and an average loss of 0.59%. It beat the 5-bar baseline only in Bitcoin.
In separate account simulations, QQQ gained about 50, Bitcoin gained 30, SPY lost 9, Ethereum lost 56. But compared to baseline, bullish engulfing only beat it in QQQ.
Only 10 qualifying occurrences, 8 finished positive, average return 1.56%. It was the only pattern above the 5-bar baseline in all four instruments.
Only 3 examples in SPY, 5 in QQQ, 1 in Bitcoin, 1 in Ethereum. In QQQ, the margin over baseline was just 0.04 percentage points. Promising observation, not proven edge.
Win rate alone can fool you. Same pattern behaves differently across markets. Headlines need rules and benchmarks underneath. Candlesticks can ring the doorbell but shouldn't sign the mortgage.
The three white soldiers pattern showed the most promise, beating the baseline in all four markets, but with very few occurrences and thin margins. The video emphasizes that win rate alone is misleading and that traders need rigorous testing with clear rules and benchmarks before trusting any pattern.
Mentioned in this Video
💡 Key Takeaways
Win rate alone is misleading
The hammer won 50.2% of the time but still lost money on average, showing that win rate without considering loss size is meaningless.
00:38Same pattern, different markets
Bullish engulfing gained in QQQ and Bitcoin but lost in SPY and Ethereum, proving patterns are not universal.
01:36Three white soldiers as the only winner
It was the only pattern above baseline in all four instruments, but with only 10 occurrences, the sample is tiny.
02:13Candlesticks ring the doorbell, not sign the mortgage
A memorable metaphor that candlestick patterns are signals, not guarantees, and should not be over-trusted.
03:13Full Transcript
[00:00] Four famous candlestick patterns. 1,011 signals. Only one finished above its market's normal 5-bar return in all four tests. The catch? Bitcoin and Ethereum gave that winner just one appearance each.
[00:13] Somebody ordered a championship belt before checking the attendance sheet. Welcome to Prove It. Ring the bell! We tested daily candles in SPY, QQQ, Bitcoin, and Ethereum starting in 2018.
[00:25] Wait for the pattern to finish, enter at the next open, Hold 5 complete bars, then exit. Commission 0.1% each way. Slippage is not simulated. These are exact program shapes, on their own.
[00:38] Not every way traders use candlesticks. First up, the hammer. Long lower wick. Dramatic bounce. Main character energy. It appeared 241 times and won 50.2%.
[00:50] A coin flip wearing a tiny construction helmet. But the average result lost 0.71%. Winning half the time does not help when the losses hit harder than the wins. Here is one actual example with the completed pattern, next open entry, and five bar exit marked.
[01:07] One picture can look convincing. That is why we count all the eligible occurrences instead of picking a screenshot where the arrow looks like a genius. The morning star promises a fresh beginning. Sunrise,
[01:20] birds singing, your portfolio finally texting you back. The test brought an alarm clock, 195 appearances, a 44.1% win rate, an average loss of 0.59%.
[01:36] It beat the ordinary 5 bar baseline only in Bitcoin. Sunrise has been rescheduled. Bullish engulfing the red candle then a bigger green body You have seen that thumbnail In separate account simulations QQQ gained about 50 Bitcoin gained 30 SPY lost 9 Ethereum lost 56 Same shape
[01:57] Four very different group chats. These account tests take one position at a time. They cannot take every overlapping signal. Here is the distinction. We counted every eligible pattern occurrence, even when 5-bar windows overlapped. Then we compared those
[02:13] average returns with each market's ordinary 5-bar average. Bullish engulfing finished above that baseline only in QQQ. Making money in a rising market and finding something beyond its normal drift are different claims. Three white soldiers marches in. Three strong green candles with extra
[02:32] body and wick rules. Ten qualifying occurrences. Eight finished positive. Average return 1.56%. It was the only tested pattern above the 5-bar baseline in all four instruments.
[02:44] Cue the victory music! Quietly. We still need to read the fine print. Three examples in SPY. Five in QQQ. One in Bitcoin. One in Ethereum. In QQQ, the average margin over the baseline was just 0.04 percentage points.
[03:01] That is a promising historical observation, not a proven universal edge. Being selective is a rule. Being reliable is something we still have to establish. Here is what survives the fight.
[03:13] Win rate alone can fool you. The same pattern can behave differently across markets. And a headline needs the rules and the benchmark underneath it. A candlestick can ring the doorbell. It should not sign the mortgage.
[03:26] Next, the 9EMA enters the courtroom. Is it finding trends or just keeping the commission department busy? We do not worship indicators. We put them on trial.