You’re Backtesting Trading Strategies WRONG — Transcript

Full transcript

1. 0:00Most of you are backtesting trading
2. 0:01strategies completely wrong. And I don't
3. 0:03mean you're calculating something
4. 0:04incorrectly. I mean the backtest itself
5. 0:07might be giving you perfectly accurate
6. 0:09numbers that are completely useless to
7. 0:11you. In this video, I'm going to show
8. 0:12you what I mean. Because this is
9. 0:13something that I see traders doing all
10. 0:15the time. You find an indicator and
11. 0:17maybe it gives you good buy and sell
12. 0:18signals. You put it up on TradingView,
13. 0:20you scroll backwards, and then you start
14. 0:23counting. Win, loss, win, win, win,
15. 0:25loss, loss, loss. And maybe you do this
16. 0:27100 times, maybe 200 times. Maybe you're
17. 0:30really dedicated and you have a full 8
18. 0:33hours to dedicate to this and you do 500
19. 0:35or 1,000 backtested trades. And then you
20. 0:37get to the end and you say, "Okay, 57%
21. 0:40win rate, this strategy works." Or 42%,
22. 0:43this thing is garbage. But there's a
23. 0:44pretty massive problem with doing that.
24. 0:47You're assuming that every single one of
25. 0:48those trades was actually available to
26. 0:51you. And it probably wasn't. As you can
27. 0:53see right here, I've got an indicator on
28. 0:54the chart. And for this example, I
29. 0:56backtested hundreds of trades. We're not
30. 0:58concerned with what the indicator is.
31. 1:00You could do this on any buy-sell
32. 1:01indicator. We're concerned with the
33. 1:03signals. We've got buys, we've got
34. 1:05sells. And I've put a position tool on
35. 1:07every single one of these trades,
36. 1:09meaning that every time the signal
37. 1:11prints, position tool, just like you
38. 1:13would do in a normal backtest. So, let's
39. 1:15just take the signals exactly as they
40. 1:16appear. No discretion, no
41. 1:18cherry-picking, every single signal. And
42. 1:21I'm going to run this across roughly the
43. 1:23last month of price action. Now, when we
44. 1:25do that, we get these results. We've got
45. 1:27182 trades. We have 89 winners and 93
46. 1:32losers. That's a 48.9%
47. 1:34win rate, profit factor of 0.95.
48. 1:37Now, normally, this is where somebody
49. 1:39decides whether the strategy is good or
50. 1:41bad. But I want you to look at
51. 1:42something. When did these trades
52. 1:44actually happen? Because I'm sitting
53. 1:45here in Europe, and let's say
54. 1:47realistically, I trade from 9:00 in the
55. 1:50morning until noon, 3 hours. That's my
56. 1:52trading window. Maybe I've got work
57. 1:54after that. Maybe I have other tasks
58. 1:56after that. Maybe I got to do some
59. 1:58gardening or mow the lawn. Maybe I got
60. 2:00kids. Maybe I want to go to the gym. It
61. 2:02doesn't matter. The point is I'm not
62. 2:04sitting here in front of this computer
63. 2:0524 hours a day. So, why the hell am I
64. 2:07including signals that happen 2:00 in
65. 2:09the morning in my strategy and my back
66. 2:12test? I wasn't awake. I couldn't take
67. 2:14it. It's not a trade. At least it's not
68. 2:16a trade for me. And this sounds
69. 2:17incredibly obvious when I say it out
70. 2:19loud. But, think about how most people
71. 2:21actually back test. They scroll
72. 2:23backwards through the chart test
73. 2:25everything they see. Tokyo session,
74. 2:27London session, New York session, middle
75. 2:29of the night, middle of the afternoon.
76. 2:31It doesn't matter. Every single signal
77. 2:33gets thrown into the same bucket. And
78. 2:35then they use that number to decide
79. 2:37whether the strategy works. So, let's
80. 2:39change one thing. I'm not changing the
81. 2:41indicator. I'm not changing the entry.
82. 2:43I'm not changing the stop losses. I'm
83. 2:45just filtering the results to give me
84. 2:48the trades during my time window. Let's
85. 2:50say 9:00 a.m. till noon. That's it. Now
86. 2:53that I filtered out every single one of
87. 2:54those trades that were not inside of my
88. 2:57personal trading window where I'm
89. 2:59actually on the charts, we can see the
90. 3:01actual numbers of this specific
91. 3:03strategy. Now we only have 51 trades. 23
92. 3:06of them were winners. That gives us a
93. 3:0845.1%
94. 3:10win rate and a profit factor of 0.81.
95. 3:13And these are the numbers that matter
96. 3:15more to me. Even if the results were
97. 3:17better, that's not the thing that I'm
98. 3:19focusing on. Filtering the time doesn't
99. 3:21magically make the bad strategy
100. 3:23profitable. The results could improve.
101. 3:25They could get worse. But, at least now
102. 3:27I'm testing something I can actually
103. 3:29execute on my charts while I'm trading.
104. 3:32Because imagine we discovered that this
105. 3:33strategy performs amazingly between
106. 3:36midnight and 4:00 a.m. but terribly
107. 3:39between 9:00 a.m. and noon. For someone
108. 3:41who can only trade between 9:00 a.m.
109. 3:43until noon, who cares what the overall
110. 3:46profitability of this strategy is? You
111. 3:48can't trade the profitable part. And the
112. 3:50opposite can happen, too. Maybe the
113. 3:52strategy looks mediocre across the
114. 3:54entire day, but during the 3 hours
115. 3:56you're actually available, it performs
116. 3:58significantly better. If you only looked
117. 4:01at the overall backtest results, you
118. 4:03might throw away a strategy that
119. 4:05actually fits you incredibly well. And
120. 4:07this gets even more important depending
121. 4:09on where you live. If you live in
122. 4:10Europe, maybe you're interested in
123. 4:12London session. If you're in the United
124. 4:13States, maybe London session happens
125. 4:15while you're asleep. Maybe New York is
126. 4:17your window. Or maybe you're a night owl
127. 4:19and Tokyo session is best for you. And
128. 4:21it's not just geography. Your life
129. 4:23matters. Maybe you're working a normal
130. 4:25job, and you can only trade from 6:00
131. 4:27p.m. to 8:00 p.m. after your normal job.
132. 4:30Then it becomes, which window do I test
133. 4:33based on my specific life? And there's
134. 4:35another layer to this. Different parts
135. 4:38of the trading day don't behave exactly
136. 4:40the same. Liquidity changes. Volatility
137. 4:42changes. Different markets open and
138. 4:44close. News tends to happen at specific
139. 4:47times of day. And when you combine every
140. 4:49hour of the day into one giant backtest,
141. 4:52you can unintentionally hide some
142. 4:54incredibly useful information inside of
143. 4:56the average. Let's just make this even
144. 4:58more obvious. Instead of testing from
145. 5:019:00 a.m. until 12:00, let's just test
146. 5:03another window. I decided to shift it to
147. 5:06the after work availability, the 6:00
148. 5:08p.m. until 9:00 p.m. Now we have, over
149. 5:11the last month, 11 trades, six winners,
150. 5:14and five losers, giving me a win rate of
151. 5:1755%. And now you can see a clearer
152. 5:20picture, that depending on what
153. 5:22indicator you're using, depending on
154. 5:23what time you're actually available to
155. 5:25sit down and trade on the charts, you
156. 5:27might need to wake up early, you might
157. 5:29need to stay up late. I'm not looking
158. 5:31for the specific results that create the
159. 5:33prettiest window, because I don't want
160. 5:35you ruining a night's sleep, because
161. 5:37sleep is incredibly important for the
162. 5:39human body. So I don't want you to go on
163. 5:41the charts from 2:00 a.m. to 4:00 a.m.
164. 5:43You're going to be a cranky individual.
165. 5:45I'm trying to let you know that your
166. 5:47life during waking hours, if you've
167. 5:50chosen day trading as your career path,
168. 5:52must be a little bit more malleable.
169. 5:54Price action doesn't matter that you're
170. 5:55only available at this time. You need to
171. 5:57look at the strategy you want to test
172. 5:59and the strategy that you want to trade.
173. 6:01You don't look for that 90% win rate
174. 6:03strategy. You look for something that
175. 6:05gives you decent consistent results
176. 6:07month over month over month and you can
177. 6:09dedicate that time window. A time window
178. 6:12where there is volatility, where there
179. 6:13is liquidity in the markets, where the
180. 6:15price will move in your direction, not
181. 6:17in days, but in minutes if you're a
182. 6:19scalper. Because the strategy needs to
183. 6:21fit your life. You shouldn't have to
184. 6:23redesign your entire life around the
185. 6:25strategy, especially if it means losing
186. 6:27sleep. This is one of the reasons why I
187. 6:29think traders massively underestimate
188. 6:32what proper backtesting actually means.
189. 6:34It's not just did the signal win or
190. 6:36lose. You need to recreate the
191. 6:38conditions under which you would
192. 6:40actually have traded. Your trading
193. 6:41hours, your spread, your commission,
194. 6:43your stop loss, your take profit, your
195. 6:46risk, potential news, and most
196. 6:48importantly, rules that you decide
197. 6:50before you see the result. Otherwise,
198. 6:52it's incredibly easy to accidentally
199. 6:54build a strategy around the past. Now,
200. 6:56there's one obvious exception to
201. 6:57everything I just said, and that's
202. 6:59automation. Because if a strategy can
203. 7:01execute completely automatically, then
204. 7:03suddenly it doesn't matter whether the
205. 7:05signal appears at 2:00 a.m. You can be
206. 7:07asleep. Your computer doesn't care. And
207. 7:09that's something I'm actually working on
208. 7:10right now. But that's a discussion for
209. 7:12another video. For manual trading, I
210. 7:14want you to try something. Go back to
211. 7:16whatever strategy you're currently
212. 7:17testing. Don't change the strategy.
213. 7:19Don't change the indicator. Don't change
214. 7:21the stop loss and don't change the
215. 7:23target. Just write down the hours that
216. 7:25you could realistically sit in front of
217. 7:26your computer and trade and backtest
218. 7:28only those hours. And compare those
219. 7:31results to your previous backtest. You
220. 7:33might discover the strategy is better
221. 7:35than you thought, or you might discover
222. 7:37it's significantly worse. But either
223. 7:39way, you've learned something far more
224. 7:41useful than the overall win rate. You've
225. 7:43learned whether the strategy actually
226. 7:45fits you. And ultimately, that's what a
227. 7:46backtest is supposed to tell you. Not
228. 7:48whether somebody somewhere could have
229. 7:50made money trading every signal on that
230. 7:52chart. Whether you, following your rules
231. 7:55during your trading window, could
232. 7:56realistically have executed it. Because
233. 7:58a trade you would have slept through
234. 8:00isn't really a part of your strategy. If
235. 8:01you guys enjoyed this video, please
236. 8:03share it with your friends. Thank you so
237. 8:04much for watching and keep an eye out
238. 8:06for the next automation video.