Radar Tape

1. The stop did not hold.

The rule was: cut every loss at −15%. Across 15 live stop-outs the tape realized a median of −33%, an average of −43%, and a worst fill of −96%. Not one exit landed at −15%.

The reason is plain: a coin can fall most of its value between the moment you look at the price and the moment you sell. This classroom book checks the price about every 10 seconds and still could not get out at its stop.

2. That is what broke the model.

This kind of trading only works if the losses stay small and the rare winner is large. When the average loss is −43% instead of −15%, no realistic number of winners pays for the losers.

The book went from a fake $100 to $26.48 across 17 fills while proving it.

3. The first backtest was wrong, and here is exactly how.

It assumed that when a price touched your stop, you sold at your stop. Re-run with a truthful assumption — you sell at the next price you can actually see — 0 of 216 tested configurations stayed profitable.

The same rule scores 7.41 under the flattering assumption, 1.21 under the honest one, and 0.12 once its three luckiest coins are set aside.

4. The screen does not pick winners either.

Tested on 115 graduated coins, no filter rule separated winners from losers better than a coin flip. The full screen would have passed 1 coin out of 115 and rejected all 8 of the biggest winners.

The filter here is a safety filter. It avoids clone tickers, offensive names, dust prints, and coins too thin to sell back out of. Nothing more.

5. What the tape is for now.

A worked example of testing a trading idea honestly. The ledger below shows every fill, including the −96% one.