Thomas
Region: US
Friday 05 June 2026 20:21:42 GMT
32738
2188
84
91
Music
Download
Comments
ً :
Lot of people r pretending to be u and selling courses
2026-06-05 20:32:03
79
Alex📉📉📉 :
How can you make sure you aren’t over fitting
2026-06-05 20:36:01
1
jrijemen :
Overfitting is a myth
2026-06-05 22:20:11
4
Isaac 🇨🇦🍁 :
In other news, water is wet
2026-06-05 21:38:14
12
Dinesh Chetty :
not everyone is overfitting. some of us use so few parameters it very difficult to overfit. the more parameters, the more chance you are overfitting. keep it no more than two.
2026-06-05 20:47:39
7
oceanwtter :
What if its overfit but over 50 years of data
2026-06-11 03:56:30
0
FS1 :
Sharpe ratio of 3 would be a pretty big red flag right?
2026-07-01 10:46:24
0
jestre :
Alright but if I for example already did OOS and IS optmizations and WFO too
2026-06-06 10:24:27
1
Dr. Memes :
Exactly! I built a trading execution system which I have set with MBO from bookmap and Back end data from Unusual whales. My test set is 150 rows and then I put it on LIVE data in the Sandbox. I do not reiterate the test set. The problem is that my test set is only 6 weeks old and MBO 1-sec for the top 20 S&P500 firms 3 months set is extremely expensive about $6K. I don't think ratios are really worth calibrating at this point because my test set is small. I am using a LinReg forward looking machine learning model. It looks at a large number of data points but it's 6 weeks sample size which is way too small. Your videos are very informative and I would like you to look at my model and suggest anything I have to do differently.
2026-06-14 23:22:52
0
Ivan :
Spot on. Genetic algos don't find the alpha, they just memorize historical noise. Those "50% return" guru backtests are literally just curve-fitted to a 10 year bull market. Without strict out of sample testing and walk forward analysis, it's just expensive historical fiction.
2026-06-07 00:44:49
1
y :
whats your take on auction market theory
2026-06-06 11:30:15
2
Noodle :
I trained my model on its test set and then tested out of sample for a 4 year (2 years of untested data) period for the first time and the equity curve looked pretty nice consistently but still not the best sharpe ratio since its breakout reversal. Unfortunately it only takes 100 trades per year and im kind of pessimistic on the thought that it only taking 1-2 trades per week is messing the the robustness of it to make sure it isnt luck. it still does outperform and returns 83% per year, Would love to get insight on this with what you think. love ur vids ty thomas
2026-07-06 07:07:29
0
meldeo :
Hey thomas do you teach all these terms and information on your youtube?
2026-07-08 05:53:57
0
Scott Worrell :
Very good information
2026-06-05 20:54:58
4
adishaz :
bro the problem is not the overfit its the thought that a one can build a HOLC strategy and expect an edge
2026-06-06 14:57:01
1
@Odiephd :
Been there done that. Lesson learned!!
2026-06-06 13:22:13
2
Exotic thoughts :
Do you like build alpha
2026-06-05 20:44:12
2
Nomel :
what if I got positive expectancy on the first run with a high sample size? not overfitted right?
2026-06-05 20:47:18
2
Osybear :
I am using ninjatrader to forward test, when would I ever consider running it on an evaluation?
2026-06-05 20:41:51
1
Carter Cole :
I feel attacked
2026-07-14 02:39:25
0
404 Not Found :
cpcv backtest
2026-07-17 21:26:14
0
Fei :
eyes already glazed over by the 4th word....voo and chill for me dawg
2026-07-03 21:45:23
0
chailodamailo :
Where is your long term broker statement and third party audit?
2026-06-11 08:39:40
0
whatever consciousness is :
I like the way you combine science with money
2026-07-08 12:00:59
0
To see more videos from user @deltatrendtrading, please go to the Tikwm
homepage.
