SNG statistics.. what's should be aimed for

[HokieHustler]

I just recently started using pokertracker and was wondering what was a good goal for average finish and % in the money as far as single table tourneys.

Thanks

[Jurollo]

how bout this?
whatever is profitable /images/graemlins/grin.gif

[AleoMagus]

As far as sngs are concerned, most of pokertrackers stats are useless (My Opinion).

That said, any ITM rate of 35-50% is good so long as you are getting a positive ROI
ROI is really what matters most though higher ITM does reduce variance. You also want at least a normal distribution of money finishes between 1,2,3rd places with hopefully more 1sts than anything else

I no longer use pokertracker and instead just plug all my results into an excel spreadsheet so that I can calculate other useful stats

Standard deviation per tourney
Total Standard deviation
Confidence in avg profit per tourney
Risk of Ruin
Finish Ratios
% of making top 3 with 4 left
% of making top 3 with 6 left
etc...

These stats can actually tell you something about leaks in your game and about how confident you should be in your results. After all, who cares if your average finish is 4.2. What does that mean anyways?
You might be getting 4th all the time and that's no good.

A better thing to know would be how your % of making the top 3 (or two or whatever) compares at any given point in the tourney to what the statistical norm would be.

Pokertrackers note taking ability is still useful so you can see who among your competition is actually finishing well but I actually don't use pokertracker at all anymore

Regards
Brad S

[MercTec]

Hi Brad,

I've read a bunch of your posts on SNGs....good stuff. About this spreadsheet. Are there formulas you use in excel? I basically suck at excel other than really basic stuff. Do you have a copy of that spreadsheet with the formulas that I could have a look at? Any help would be appreciated. Thanks!

[Hood]

What stats do you take so you can calculate '% of making top 3 with 4 left' etc? I just track Tounry length, profit of each tourney and finish, from which I can calculate average length, average finish, ITM and ROI, but that's about it.

[AleoMagus]

Hi,

Here is a sample spreadsheet that I made for recording and analyzing sng results

Spreadsheet (http://www.aleomagus.freeservers.com/spreadsheet)

Actually, it isn't exactly what I use anymore as I am almost constantly modifying it whenever I get bored and have nothing better to do.

The results listed are just sample results. I just made them up

This spreadsheet is bad for a few reasons. It is only designed to input from 20+2. This is ok in itself, but I'm making another where you will fill in one feild with the sng buy-in and it will change the entire spreadsheet to suit that sng type. I'm playing a lot of 30+3 now, and a universal spreadsheet seems like a better idea than constantly modifying whenever I move up.

If a person wanted to use this spreadsheet for another buy-in, there is a lot of fiddling with individual formulas, etc... that would need to be done.

Remember, the only fields that should ever be filled in are the grey fields. The rest all calculates automatically. Actually, the # of players feild can be changed too for all you pokerstars players. I just have it set at 10 becasue I'm only playing on Party lately.

If you ever went beyond the 100 sng mark (which is all I've included in this sample), you would need to copy and paste the results rows for as many more as you would need. You would also need to change the source data entries for the graphs. (Not hard stuff to do, but a pain nonetheless)

This spreadsheet also does not track length of tourneys so it will not give an hourly rate. These days I only play about 50 sngs/mo so I never really care much about that, but for some players it is very important, particularly when thinking about multi-tabling etc...

Excel looks daunting at first, but it isn't. I just started learning about it a few months ago. I suspect that in another year, I will have a sng spreadsheet that will make this one look silly.

There are a lot of other interesting stats which I want my spreadsheet to record
Things like:

Confidence in each finish %
Streak/Slump stats
Risk of Ruin based on Current online BR
Stats of results following wins/bubble finishes, etc...
Lots more graphs (what can I say, I like graphs)
Results when multi-tabling
Others stuff

I suspect that there are other players (Bozeman Maybe. He always seems to have nice graphs in his results posts) who have really good spreadsheets also.

Hope this helps
Brad S

[AleoMagus]

For % of making top 3 with 4 left, take your # of 1,2 & 3rd place finishes and divide by 1,2,3, & 4th place finishes

This value should be higher than .75. If it is a lot lower, then your bubble play is bad and you might be getting to the final 4 shortstacked too often. If it is a lot higher, I envy you.

Similar things can be done to calculate any finish ratio

Check out my spreadsheet sample. It does it all automatically

Regards
Brad S

[Hood]

Having problems downloading the sample, permission denied

[AleoMagus]

Anyone else having this problem?

Seems like it works to me, but then - it's my site so...
Did you pick 'Run from current location'? I'm not sure how that would make any difference but that is what I did and it worked.

Regards
Brad S

[Hood]

[ QUOTE ]
Anyone else having this problem?

Seems like it works to me, but then - it's my site so...
Did you pick 'Run from current location'? I'm not sure how that would make any difference but that is what I did and it worked.

Regards
Brad S

[/ QUOTE ]

Not sure what you mean by 'run from current location'... I clicked the link then on the file, and get the following:

Forbidden
Remote Host: [xx.xx.xxx.xxx]

You do not have permission to access http://www.aleomagus.freeservers.com/Spreadsheet/spreadsheet%20sample.xls
Data files must be stored on the same site they are linked from.

Thank you for using FreeServers

[BradleyT]

This Page (http://www.aleomagus.freeservers.com/Spreadsheet/) should take you to a directory structure and then you can click the spreadsheet to download it.

When you directly request the spreadsheet it notices that you didn't click any link within his domain (by checking the URL Referrer) and says, hey you can't do that. Basically it's a safeguard so that he couldn't show images on THIS site but have them hosted on HIS site which would use up bandwidth from his site.

[Jsb]

worked fine for me, but i opened as is and then saved it from there instead of saving directly. no particular reason why, just habit i suppose. but it worked, so yeah.
very impressive, by the way. very impressive.
i'm rather familiar with excel, i've taken a few physics classes where we used it extensively, i just didn't have much of an idea what exactly to calculate, what data would be useful to analyze and what not. this is very helpful though, thanks a bunch!

[Hood]

Ok managed to download it here at work, must be my firewall or something at home.

It looks *very* useful, thanks a lot. I think my main problem was not having position columns 1-10 for each result - this makes many more calculations possible (or at least easier).

The only bit that confuses me is the 'confidence' calculations. Do the confidence % equate to anything I can relate to? E.g. a win confidence of 95.4% - does that mean a I can be 95.4% confident that my total # of wins is accurate enough to rely on?

[AleoMagus]

The confidence calculations show the degree of confidence that we can have in our average profit per tourney.

The three confidence calculations I include in that sample are

+/- $1/tourney
+/- $2/tourney

and what I call 'winning' confidence which is +/- the actual amount that you are making per tourney.

This is to say that if I make $7 per tourney on average and my 'winning' confidence is 95%, then I can be 95% sure that I make $7 +/- $7. Another way of saying this would be to say that I can be 95% confident that I make between $0 and $14 per tourney

Obviously, this is not precise at all, but I still think 'winning' confidence is important because establishing to yourself that you are capable of at least breaking even is crucial.

In order to get a high degree of confidence in results to +/- $2, or +/1 takes a lot longer.

For the average player (with average standard deviation) you will need about 300 tourneys to even get a 60-70% confidence in results to +/- 10%.

If you fiddle with the number of tourneys played, you can see how the confidence intervals increase over time. It's actually pretty humbling to see how little confidence we can have over what seem like large samples.

The Z values which are listed above the confidence calculations are really just a part of the calculation and I only have them listed to make the calculation easier for me. They don't mean anything on their own.

As far as really low confidence results are concerned, they don't mean much either. Anything below 50% is like saying 'No certainty whatsoever about the preciseness of results in that range'.

Regards
Brad S

I'm no statistician actually and I'd appreciate any expert feedback on my calculations and if I am interpreting the results correctly.

[Hood]

Thanks for the info.

I think I be moving my data across to this format. I think I'll try and modify it for a few things: firstly so it supports different SNG buyins (instead of hard-coding in the buyin amount and amount won for each place I'll reference it from a column, and use a sheet for each SNG buyin). Also mod it to record the time each tounry took.

Thanks again for all the help.

[AleoMagus]

I'm actually going to post a revised (universal) version soon

I'm adding a few more stats and implementing a hourly wage calculation also

because I hate keeping track of how long each sng takes, I am planning on some approximation which I think will work ok.

The spreadsheet will simply fix a previously alloted 'average' time to each finish and will automatically calculate an hourly wage this way

For example

1st - 1hr
2nd - 1hr
3rd - 50 min
4th - 40 min
5th - 35 min
6th - 30 min
7th - 25 min
8th - 20 min
9th - 15 min
10th - 10 min

(this is just a possibility, maybe too much time per finish here)

I know not every finish takes the same amount of time, but in the long run, it will form an average and this will make a good approximation with zero effort. Besides, a person needs to take into account waiting times, game selection times, etc... if they want to be really precise so this may not be far off anyways.

I'm also trying to implement a Day of week analysis to detect ROI, ITM, etc... for each day of the week. This could even be done for morning/afternoon/evening/night.

The problem is, what are already relatively small sng samples become even smaller when you chop them up and it is almost meaningless to infer too much from (say) a 30 sng tuesday evening sample.
This would really only become useful after a HUGE total sample (or if you played at a few specific times regularly).

Anyways, I'll post another sample spreadsheet in a separate thread sometime soon.

Regards
Brad S