Ok very interesting things going on with the system, unfortunately again have no official plays which is where the system is undefeated but I have a couple experimental plays for it to try to get some volume going on its plays as 3 weeks in a row of no official picks is unacceptable after having official plays for 3-4 weeks in a row. This week is crazy with so many fighters having a lack of ufc history so I had to adapt the system to use non-ufc history to get the 2 experimental plays.
2 very funny things going on first of all in the Allen fight night thread I posted 8 picks ranked in order of best to worst based on what the system thinks and the top 4 plays all lost bottom 4 all won... so being contrarian seems to be where the system succeeds including when it chose Brady hiestand +300 the week that everyone was choosing Nakamura... anything over +20 edge has been pretty profitable as well overall despite the RDA loss which even he looked great rd 1 just too old and inactive.
Also not only did the bottom 4 win but they were in reverse of what would've been the strongest bets. Like Price +136 and what's his name Tazari? That's not right name... Wellmakers opponent +176 were supposed to be the worst but they were the best picks
Other thing is I'm working on a couple new versions of the system trained on entire paramount era except the last 5 cards and tested on the 5 most recent cards to demonstrate that it works on completely unseen cards. And super funny the first one goes 2-10 on those 5 cards for a nearly 6 unit profit if you fade it and +50% ROI and very consistent profit on every card. So we're perfecting how to lose reliably on that version ?? hahaha
So this is exciting this week we will be testing the originals 2 experimental plays and whatever the 2 new versions come up with. We also have a lot of big edges on non plays this week to see if the success continues on over +20 edges
On a quest to be the very best
To remove first post, remove entire topic.
Ok very interesting things going on with the system, unfortunately again have no official plays which is where the system is undefeated but I have a couple experimental plays for it to try to get some volume going on its plays as 3 weeks in a row of no official picks is unacceptable after having official plays for 3-4 weeks in a row. This week is crazy with so many fighters having a lack of ufc history so I had to adapt the system to use non-ufc history to get the 2 experimental plays.
2 very funny things going on first of all in the Allen fight night thread I posted 8 picks ranked in order of best to worst based on what the system thinks and the top 4 plays all lost bottom 4 all won... so being contrarian seems to be where the system succeeds including when it chose Brady hiestand +300 the week that everyone was choosing Nakamura... anything over +20 edge has been pretty profitable as well overall despite the RDA loss which even he looked great rd 1 just too old and inactive.
Also not only did the bottom 4 win but they were in reverse of what would've been the strongest bets. Like Price +136 and what's his name Tazari? That's not right name... Wellmakers opponent +176 were supposed to be the worst but they were the best picks
Other thing is I'm working on a couple new versions of the system trained on entire paramount era except the last 5 cards and tested on the 5 most recent cards to demonstrate that it works on completely unseen cards. And super funny the first one goes 2-10 on those 5 cards for a nearly 6 unit profit if you fade it and +50% ROI and very consistent profit on every card. So we're perfecting how to lose reliably on that version ?? hahaha
So this is exciting this week we will be testing the originals 2 experimental plays and whatever the 2 new versions come up with. We also have a lot of big edges on non plays this week to see if the success continues on over +20 edges
This is the normal edge chart like I've posted every week... each line is the fighter system chose and E means experience reliability D means damage reliability
No picks for this version but keep an eye on the +20ish or higher edges
This is the normal edge chart like I've posted every week... each line is the fighter system chose and E means experience reliability D means damage reliability
No picks for this version but keep an eye on the +20ish or higher edges
Now here's the 2-10 on 5 card events +nearly 6 units +50% ROI fade system
I have 3 versions balanced made +nearly 8 units on same 5 cards and max profit made nearly 10 units on same 5... every version profited on all 5 cards consistently
All 3 versions fade Mike Malott to make Buckley +118 an official play and the only play this card
UFCC2.1 FADES — OCT 17 -------------------------------- Model Fade Pick Odds Original Buckley +118 Balanced Buckley +118 Max Profit Buckley +118 -------------------------------- Classifier originally picks: Mike Malott -140 Model Malott win: 65.2% Model Malott edge: +9.2% Experience rel: 50% Damage rel: 70.4%
On a quest to be the very best
Now here's the 2-10 on 5 card events +nearly 6 units +50% ROI fade system
I have 3 versions balanced made +nearly 8 units on same 5 cards and max profit made nearly 10 units on same 5... every version profited on all 5 cards consistently
All 3 versions fade Mike Malott to make Buckley +118 an official play and the only play this card
UFCC2.1 FADES — OCT 17 -------------------------------- Model Fade Pick Odds Original Buckley +118 Balanced Buckley +118 Max Profit Buckley +118 -------------------------------- Classifier originally picks: Mike Malott -140 Model Malott win: 65.2% Model Malott edge: +9.2% Experience rel: 50% Damage rel: 70.4%
Finally heres the 2-10 system's fade picks value chart... the probabilities are low because it thinks the other side has the higher chance to win but we're fading its higher chances cause fading this system is what is supposed to be profitable
Finally heres the 2-10 system's fade picks value chart... the probabilities are low because it thinks the other side has the higher chance to win but we're fading its higher chances cause fading this system is what is supposed to be profitable
Hot Forum Topics
