Spent two days upgrading and back testing model... Starting new 10 day trial...

[usma1992]

I understand your thought process. Right this second it is hitting at a 99% rate for the games... I am looking at. Therefore, even at a 5% return... for every 20 bets... it returns a full unit. If you win 99 and lose 1 huge upset... you still netted 4+ units. In addition, if you increase the bet size to 10% of lets say 9% that is worth 3x of the spread bets.

Therefore, those net 4 units are really worth 12 of the other type.

Yes, there is a possibility that you get an upset or maybe many upsets. However, my system isn't just looking at the spread... it is also looking at my system above and beyond the spread. So chance of failure should become less.

Dave

[Neek762]

What I'm trying to say lol is I think if you're going the ML route, you'll need to eventually parlay hedge, or bet enough to win a unit. Instead of betting $100, you'll need to bet to WIN $100. Not sure if that makes sense

[Neek762]

Again unless the -1500's, -1000's, -800's and 1000% locks, I don't know if flat betting the same amount on ML is the best route. There's strategies in Roulette like that - you can cover a bunch of numbers to win a chip, but it takes 5-6 hits until you are safe. Get whacked once and you're out 10 chips or whatever it may be. You'll have to parlay or bet enough to win a unit. 5-10% at a time is way too high of a risk of ruin for me though. Much rather make 100 1% wagers instead of 10 10% wagers

[usma1992]

Betting:
I am betting ML as moneyline bets so they are paying -1500 or so bet 5-10% of my roll on those bets because their hit rate... is 99%+. Which means I need to win 10 bets before I lose 1 or approximately maybe 15 bets. If I team them up like the other night it generates a nice 4-5% return with minimal risk.

Spread bets and O/U bets:
I try to keep these bets to 3% of my roll. Two bets tonight.

FGCU vs. E. Kentucky is intriguing bet because my number is way off from Vegas. I am curious who is right. I have 158... they have 143 and dropping. I checked my stats and they were all correct, so I guess we will see.

Look at NBA... my system rarely picks games... and it picked two tonight.

Dave

[Neek762]

Kind of confused to how you're betting if it didn't pay anything? Let's use $100 as the bankroll example and SMU -275 as the wager. Are you wagering $5 (5%) on -275 to win $1.82, or are you wagering $13.8~ to win $5, which is a unit, in this example. If I'm wagering on + money I'll bet my unit size, if it's - money I'm betting to win my unit size. Is that what you're doing? The only way to kill those odds are to parlay it with another game and adjust the bet amount. SMU+Murray St hedged with SMU+Belmont. As long as you're confident SMU will NOT lose

[prolexus]

Thank you again.
Be well.

[usma1992]

Won ML ... but it didn't pay anything... but 4-5% feels worth it.

Time	AWAY TEAM	PROJ AWAY SCORE	HOME TEAM	PROJ HOME SCORE		TOTAL	VEGAS SPREAD	VEGAS O/U		SPREAD BET	O/U BET
8:00 PM	SMU	81	RICE	67		148	6.5	146.0		Away Minus	Too Close
9:00 PM	FLA GULF COAST	73	E KENTUCKY	85		158	-8.5	143.0		Too Close	Over

[usma1992]

Can't catch a break lately... Under was 19 points under and it went into OT and I lost money. However, I count that as a win because my system picked it in regulation. Today's bets... nothing very attractive.

Time

AWAY TEAM

PROJ AWAY SCORE

HOME TEAM

PROJ HOME SCORE

TOTAL

VEGAS SPREAD

VEGAS O/U

SPREAD BET

O/U BET

Actual Away

Actual Home

6:30 PM

DEPAUL

58

ST JOHNS

88

146

-21.5

149.0

Parlay ML

Too Close

0

0

7:00 PM

OKLAHOMA ST

53

HOUSTON

82

135

-20.5

128.5

Parlay ML

Too Close

0

0

[Neek762]

I don't think 20% bankroll risk is the smartest move unless that 1% chance of losing the wager turns into a 0% risk. I'm very strict about 1% = 1 unit with 5% being my max ... definitely not telling you what to do though, Dave!

[usma1992]

I have decided to increase my bankroll ML bet to 10% and maybe up to 20%. Based on past history my ML bets have a 1% chance of losing with at least a 5% return on your investment. Not sure if this is the right move. I need to sleep on it and figure out if it makes sense.

Dave

[usma1992]

That ending was rough... same with Quinnipiac game.... in the last second... I've lost two games in the last three days. But that is when bankroll management comes into play. 44-31... Both NBA hit yesterday easily. With tougher filters... maybe it is finally generating winners... just not that many picks.

Time

AWAY TEAM

PROJ AWAY SCORE

HOME TEAM

PROJ HOME SCORE

TOTAL

VEGAS SPREAD

VEGAS O/U

SPREAD BET

O/U BET

Actual Away

Actual Home

7:30 PM

MORGAN ST

64

NC CENTRAL

77

141

-8

148.5

Too Close

Under

0

0

8:00 PM

TENNESSEE TECH

58

MCNEESE ST

86

144

-17

136.0

Parlay ML

Too Close

0

0

[Neek762]

Yeah, ended up getting MSM before Dave released it as a pick actually, but then had to hedge live with Siena +8. I thought MSM was going to boat race Siena

[RM Logic]

Talk about a bad beat. Siena gets a layup with 1 second left to cover the spread. After MSM chokes away a 16 point lead.

[Sund]

Are you going to do baseball after college basketball is over?

[Neek762]

Been lurking for a while and finally got around to making an account. Appreciate the work! End of the day, variance will do what variance does. From what I've seen so far, what you're doing is working. So when a loss does come, I wouldn't trip off of it too much since it's impossible to hit them all anyways!

[usma1992]

So I got crushed yesterday... 1-5 on the slate... If you were following the Stetson game, they went into overtime and actually the bet covered the spread. However, as I model... I consider that a loss. I had Stetson -4.5 and they tied in regulation. There is no way to model for OT. Regardless of what happens in OT... my results are based on regulation time.

Well, I looked into Away - bets and I decided to tighten them up a little. I added an additional filter that filtered out two of the losses last night. I know that it is done after the fact. I am gearing the model to go live next year. So I am making improvements daily.

Even after adding the filters... it did produce one Away - bet today... 44-30... just below 60%.

Good Luck Dave

Time

AWAY TEAM

PROJ AWAY SCORE

HOME TEAM

PROJ HOME SCORE

TOTAL

VEGAS SPREAD

VEGAS O/U

SPREAD BET

O/U BET

Actual Away

Actual Home

2:00 PM

MOUNT ST MARYS

81

SIENA

60

140

8.5

135.5

Away Minus

Too Close

0

0

[Sund]

I’m sorry , it was Morehead st !

[usma1992]

Which game? I ran the system at midnight last night so lines probably adjusted... I was awake and decided to run it. I probably should have waited.

I might need to establish a timeframe that I run the lines. 12 on weekdays and 9AM on weekends.

I was attending a Regional wrestling tournament this morning... so I knew I wouldn't get around to it. I forgot to post at midnight... So I posted this morning.

Enjoy... Hopefully it doesn't affect the outcome.

Dave

[Sund]

Question? Wonder why the point spread dropped from 11 1/2 to 10 1/2 from the time you posted?

[usma1992]

I guess Quinnipiac had a dunk as time expired... didn't count of course. And Manhattan didn't foul with 13 seconds ... wtf... regardless...

Time	AWAY TEAM	PROJ AWAY SCORE	HOME TEAM	PROJ HOME SCORE		TOTAL	VEGAS SPREAD	VEGAS O/U		SPREAD BET	O/U BET
3:00 PM	APPALACHIAN ST	84	GA SOUTHERN	66		150	11	144.0		Away Minus	Too Close
4:00 PM	MOREHEAD ST	84	TENNESSEE TECH	59		144	11.5	134.5		Away Minus	Too Close
6:00 PM	NOTRE DAME	61	PITTSBURGH	77		138	-11.00	128.0		Home Minus	Too Close
8:00 PM	FLORIDA ST	80	LOUISVILLE	68		148	6.00	150.5		Away Minus	Too Close
10:00 PM	HAWAII	63	UC-IRVINE	76		138	-8.00	134.5		Home Minus	Too Close
4:30 PM	STETSON	83	C ARKANSAS	74		157	4.50	154.0		Away Minus	Too Close

[usma1992]

OMG... I got a push on NBA somehow at 249... and I feel like I just hit a huge parlay. I am ready for Saturday College Basketball. Still at 63% or better and that doesn't include backtested data....

The algorithms I have in place that are producing 43-25... also produced 143-37... 79%. Obviously, I know that is not realistic predicting forward... but it is over 63% and documented on this thread.

Let's see what Saturday brings...

Dave

[usma1992]

James...

First of all... as I upgrade the model... I always save it the newest version to my google drive. I therefore... can go back and test at some point what any version of the model. I agree that it has been working but I am constantly looking for weaknesses ... especially after I watch the games.

I try and figure out what I missed and can I do better. Especially when there seems to be a noticeable weakness in a certain category. I am not making major adjustments. I am not adjusting my point totals at all. All I am adjusting is which bets I place based on the point totals and additional match up criteria.

I think I got the ML Parlay piece dialed in. In the last few weeks... I have lost 3 ML Parlays. I have fine tuned the picks to only missing one game out of 65 or so...

I am now 43-25 with the 1 point loss Quinnipiac... they were up huge all game. Many Home - games have been losing lately... I count 3 recently so I tighten just that bet up today.

Bottom line... I am not revamping entire system every time.

Lost a college bet and NBA tonight... NBA is driving me nuts... I revamp and revamp and revamp... and then breakeven or lose. I am running it for another trial.

Anyway, thank you for the feedback... Is there another thread worthwhile that i should read? For insite...

Dave

[tecmoslayer]

The only pick I missed today. Qunnipiac can't hang onto big leads.

Time	AWAY TEAM	PROJ AWAY SCORE	HOME TEAM	PROJ HOME SCORE		TOTAL	VEGAS SPREAD	VEGAS O/U		SPREAD BET	O/U BET
7:00 PM	QUINNIPIAC	82	MANHATTAN	70		152	7	152.5		Away Minus	Too Close

[jamesrav]

Time	AWAY TEAM	PROJ AWAY SCORE	HOME TEAM	PROJ HOME SCORE		TOTAL	VEGAS SPREAD	VEGAS O/U		SPREAD BET	O/U BET
7:00 PM	QUINNIPIAC	82	MANHATTAN	70		152	7	152.5		Away Minus	Too Close

"the proof is in the pudding" or in this case the bankroll, and it sounds like that is doing very nicely. In following the thread, my only question would be about the constant 'upgrading' - are these very minor tweaks to weights, or something more than that? I used to do the same thing with my greyhound betting, after a bad stretch I'd go back and 'adjust' things to improve what would have happened. But that's overfitting data isn't it? At some point isn't a model / algo the same as a pizza joint: once the pizza is doing well and the customers are happy, you don't touch the recipe. If they do poorly one day (or week), maybe it was due to the weather, or other circumstances unrelated to the pizza. 41/63 on spread bets is well above what's needed to make a lot of money, so if you can keep things at that level over time you've done something quite spectacular.

[usma1992]

Time	AWAY TEAM	PROJ AWAY SCORE	HOME TEAM	PROJ HOME SCORE		TOTAL	VEGAS SPREAD	VEGAS O/U		SPREAD BET	O/U BET
7:00 PM	QUINNIPIAC	82	MANHATTAN	70		152	7	152.5		Away Minus	Too Close

[bubba99]

So I spent the afternoon on NCAAB Parlay ML bets... I was working with the system over and over to see if I could find a sweet spot. Two of my ML bets have caused upsets in my parlays... in the recent past and I was attempting to eliminate those and keep the ones that were working.

The two upsets against my ML were Arizona/Oregon St.... by 3 and New Mexico last night. Those cost me.

So I worked hard and narrowed it down to a reasonable criteria. Back tested I am 100% over the life of 57 games. However, these games are going to -1000 and above. But that is still 10% on your money. And if you team them up in a parlay... the payout increases obviously.

Scenario 1: 77-1 loss. So assuming for every ten wins... one loss cancels your net gain.. That is still 6-1 on your money. Breakeven is basically 7 losses over the course of the 77 games. But again these are ML and no game is less than a 11 point favorite. I also take my propreitary point total into account.

Scenario 2: 57-0. Not a single loss and this is back tested. There have been several upsets recently Duke Pitt ... Kentucky... Tennessee South Carolina... Arizona Oregon St. My system has only been affected by the Arizona game of the major upsets. These parameters eliminated that game and the New Mexico game too. I checked.

I think I a going to move forward with scenario two... I also think that I am going to increase my wager size of ML bets to 6% to 10% of my bankroll. And I am going to concentrate on Parlays when available. It isn't picking many games... so parlays may not be available.

Dave

Either way you go looks like your system is doing well, keep it up!

[usma1992]

I am appreciate you prolexus... you are one of the few that keep me posting. And I am winning... most days... Always trying to upgrade...

But ty...

Dave

[prolexus]

Your time, effort, and approach is appreciated in helping us all become successful on the NCAA College Basketball front.
cheers.

[usma1992]

So I spent the afternoon on NCAAB Parlay ML bets... I was working with the system over and over to see if I could find a sweet spot. Two of my ML bets have caused upsets in my parlays... in the recent past and I was attempting to eliminate those and keep the ones that were working.

The two upsets against my ML were Arizona/Oregon St.... by 3 and New Mexico last night. Those cost me.

So I worked hard and narrowed it down to a reasonable criteria. Back tested I am 100% over the life of 57 games. However, these games are going to -1000 and above. But that is still 10% on your money. And if you team them up in a parlay... the payout increases obviously.

Scenario 1: 77-1 loss. So assuming for every ten wins... one loss cancels your net gain.. That is still 6-1 on your money. Breakeven is basically 7 losses over the course of the 77 games. But again these are ML and no game is less than a 11 point favorite. I also take my propreitary point total into account.

Scenario 2: 57-0. Not a single loss and this is back tested. There have been several upsets recently Duke Pitt ... Kentucky... Tennessee South Carolina... Arizona Oregon St. My system has only been affected by the Arizona game of the major upsets. These parameters eliminated that game and the New Mexico game too. I checked.

I think I a going to move forward with scenario two... I also think that I am going to increase my wager size of ML bets to 6% to 10% of my bankroll. And I am going to concentrate on Parlays when available. It isn't picking many games... so parlays may not be available.

Dave

[usma1992]

Bets Tonight...

Time	AWAY TEAM	PROJ AWAY SCORE	HOME TEAM	PROJ HOME SCORE		TOTAL	VEGAS SPREAD	VEGAS O/U		SPREAD BET	O/U BET
12:00 PM	IUPUI	58	WI-GREEN BAY	77		135	-12	135.0		Parlay ML	Too Close
7:00 PM	TULANE	66	SMU	81		147	-8	159.5		Home Minus	Too Close
7:00 PM	COASTAL CAROLINA	64	JAMES MADISON	89		154	-19	155.5		Parlay ML	Too Close
7:00 PM	CLEVELAND ST	85	DETROIT	64		150	7.5	147.5		Away Minus	Too Close
10:00 PM	UC-IRVINE	78	CAL POLY-SLO	59		138	14	131.5		Parlay ML	Too Close
6:30 PM	MAINE	63	UMASS-LOWELL	75		138	-8.5	139.0		Too Close	Under
7:30 PM	BELLARMINE	70	KENNESAW ST	85		155	-9.00	152.0		Home Minus	Too Close

[usma1992]

1-2 on Spread and O/U bets... so now 41-22. Also lost ML bet because New Mexico got upset. I plan to spend a significant amount of time on the Parlay ML. I want to get that near 95% hit rate.

I will be looking closely on the Lafayette bet and the New Mexico bet. I will keep charging on.

Dave

[Sund]

Thanks! Continue your great work , much appreciated

[usma1992]

Betting:

I am treating all individual bets using 3% of my bankroll. I also then bet the ML Parlay as a separate 3% bet. It normally doesn't net very much. But when the teams are very close to sure things... Making 6% on a ML Parlay isn't bad. That is only 6% of the bet amount, not 6% of my bankroll.

The parameters are very extensive and it depends on the type of bet. I have them divided into 7 categories if you include ML parlays. First, I generate my own proprietary point totals. Then I determine where I am weak and strong based on past history. For instance on the Parlay ML bets... the spread has to be greater than 10 and my system had to side with the favored team by more than 3 or 4 points past the spread. So my system is saying the favorite will win by 14 or so.

Once I have an algorithm tested using past data... I test it moving forward. I have been running this test since Jan 2 and it is definitely working. Hope that helped. Giving NBA a try again... I loosened it up... but I like the work I put into it.

Dave

[usma1992]

Bets tonight... 40-20 right now... that isn't including backtested algorithms.

Time	AWAY TEAM	PROJ AWAY SCORE	HOME TEAM	PROJ HOME SCORE		TOTAL	VEGAS SPREAD	VEGAS O/U		SPREAD BET	O/U BET
6:00 PM	UNC-GREENSBORO	89	VMI	66		155	16.5	148.0		Parlay ML	Too Close
8:00 PM	LA-MONROE	63	LA-LAFAYETTE	80		143	-12	145.0		Parlay ML	Under
9:00 PM	UAB	61	N TEXAS	74		135	-7	133.0		Home Minus	Too Close
9:00 PM	VANDERBILT	57	AUBURN	83		139	-17.5	143.0		Parlay ML	Under
10:30 PM	BOISE ST	65	NEW MEXICO	81		146	-11	151.0		Parlay ML	Too Close

[usma1992]

Looks like everything won last night. Clemson won ML but didn't cover which is ok. Seton Hall killed by 33. Easy bet 40-20 now on single bets. I will answer questions later today when I have more time.

Dave