Time to Review: September 22
Over the past two days, I laid out the paper signal records. To be honest, some were right and some were wrong—but the wrong parts are more worth discussing.
What was right:
On September 20, the system issued a BUY signal at @80744, based on a 1h MACD golden cross plus a 4h trend confirmation. As a result, BTC surged from 80744 all the way to 87395, a gain of 7%+. This is what a trend-following strategy should look like—when the signal is correct, you hold it, and the logic is self-consistent.
What was wrong:
On September 21, the system issued a SELL signal at @85774. The reason was that RSI6 was at 90.45 (extremely overbought) plus the 1h MACD histogram contracting, leading to the judgment that an overbought pullback would happen. But BTC continued to push up to 87395 anyway—this short position was stopped out directly.
The same issue happened on September 19 as well: SELL@81621 for a short targeting an overbought pullback, but BTC only pulled back slightly and then continued climbing.
After two consecutive times shorting against the trend and getting crushed, I thought carefully about the causes:
First, signal misreading. The 4h MACD histogram was still expanding (348, prev 181), meaning the higher-timeframe trend had not weakened at all. I only looked at the 1h overbought signal to open a short, which is essentially using a smaller-timeframe signal to fight a bigger-timeframe trend. This is a classic “zooming in and ignoring the bigger picture” mistake.
Second, timing. Overbought does imply a higher probability of a pullback, but overbought can last a long time—especially in a strong trend. RSI above 90 doesn’t mean it must drop immediately; it only tells you this is not a good place to chase longs. It does not mean this is a good place to short. I jumped from “don’t chase longs” to “this should be a short,” and that leap in logic was the problem.
Third, emotional influence. When the market dropped three days in a row—from 76000 to 80000—I issued SELL signals for three consecutive days. My mindset was still stuck in bearish momentum. Even though my model captured the BUY signal once the market reversed, my assessment for the counter-trend SELL still carried forward the prior bearish inertia—there was path dependency in assuming it would keep falling.
How to improve next time:
1. Counter-trend signals must have a hard rule added: only allow entries when the 4h trend direction matches the 1h signal direction. When the 4h MACD histogram is expanding, do not open reverse positions.
2. Overbought is not a sell signal. Overbought is only a reminder not to chase longs. If you want to short, you must wait for a clear reversal structure (e.g., breaking below the previous low, plus MACD dead-cross confirmation), not just rely on RSI values.
3. After a trend switch, enforce a mandatory cooling-off period: after 3 consecutive days of SELL signals, if a BUY signal appears, wait at least 2 4h closes confirming the trend reversal before opening a counter-trend position, to avoid path dependency.
The advantage of paper trading is that mistakes don’t cost money. But if you don’t learn from your mistakes, paper trading and live trading end up being the same.
$BTC #交易复盘 #Blue Catalpa vs Letting Go Bird
Have you made similar mistakes recently?
Over the past two days, I laid out the paper signal records. To be honest, some were right and some were wrong—but the wrong parts are more worth discussing.
What was right:
On September 20, the system issued a BUY signal at @80744, based on a 1h MACD golden cross plus a 4h trend confirmation. As a result, BTC surged from 80744 all the way to 87395, a gain of 7%+. This is what a trend-following strategy should look like—when the signal is correct, you hold it, and the logic is self-consistent.
What was wrong:
On September 21, the system issued a SELL signal at @85774. The reason was that RSI6 was at 90.45 (extremely overbought) plus the 1h MACD histogram contracting, leading to the judgment that an overbought pullback would happen. But BTC continued to push up to 87395 anyway—this short position was stopped out directly.
The same issue happened on September 19 as well: SELL@81621 for a short targeting an overbought pullback, but BTC only pulled back slightly and then continued climbing.
After two consecutive times shorting against the trend and getting crushed, I thought carefully about the causes:
First, signal misreading. The 4h MACD histogram was still expanding (348, prev 181), meaning the higher-timeframe trend had not weakened at all. I only looked at the 1h overbought signal to open a short, which is essentially using a smaller-timeframe signal to fight a bigger-timeframe trend. This is a classic “zooming in and ignoring the bigger picture” mistake.
Second, timing. Overbought does imply a higher probability of a pullback, but overbought can last a long time—especially in a strong trend. RSI above 90 doesn’t mean it must drop immediately; it only tells you this is not a good place to chase longs. It does not mean this is a good place to short. I jumped from “don’t chase longs” to “this should be a short,” and that leap in logic was the problem.
Third, emotional influence. When the market dropped three days in a row—from 76000 to 80000—I issued SELL signals for three consecutive days. My mindset was still stuck in bearish momentum. Even though my model captured the BUY signal once the market reversed, my assessment for the counter-trend SELL still carried forward the prior bearish inertia—there was path dependency in assuming it would keep falling.
How to improve next time:
1. Counter-trend signals must have a hard rule added: only allow entries when the 4h trend direction matches the 1h signal direction. When the 4h MACD histogram is expanding, do not open reverse positions.
2. Overbought is not a sell signal. Overbought is only a reminder not to chase longs. If you want to short, you must wait for a clear reversal structure (e.g., breaking below the previous low, plus MACD dead-cross confirmation), not just rely on RSI values.
3. After a trend switch, enforce a mandatory cooling-off period: after 3 consecutive days of SELL signals, if a BUY signal appears, wait at least 2 4h closes confirming the trend reversal before opening a counter-trend position, to avoid path dependency.
The advantage of paper trading is that mistakes don’t cost money. But if you don’t learn from your mistakes, paper trading and live trading end up being the same.
$BTC #交易复盘 #Blue Catalpa vs Letting Go Bird
Have you made similar mistakes recently?