Maximum drawdown
 

Would I be right in saying that a "maximum drawdown" is the biggest drop from the highest point of the bank?

Ie if your bank grows from 100 to 25,000 then goes down to 12,000, is the maximum drawdown $13,000?

Also, do you express it as a %? (In this case, 52% of bank was lost from maximum down to 12,000).

If this is all true, then over an extended period, what's the maximum drawdown you want to be able to endure? For example, the system I am currently testing in another thread has, over a 2 year period, a 58% maximum drawdown. This is over 6000 bets, using 1% of bank as the bet size.

Would this be considered "acceptable"?

If I remember rightly jfc put a maximum drawdown calculator on a site for us somewhere last time we discussed this.

The answers to all your questions are "It depends" (except the first one which is "yes").
In your case representing your maximum drawdown as a percentage seems pretty sensible because your bet size is a proportion of your bank so your drawdown and your bets are in the same "units". Although, if you're reducing the size of your bets as your bank shrinks who cares what the maximum drawdown is - you should be able to take a huge drawdown without exhausting your bank - until you get to the minimum bet size of your bookie. Of course, building it up again is going to take ages with the teeny weeny bets that your miniscule bank allows (not a fan of reducing bet systems myself).
I tend to increase my bet size in the (unlikely) event that my bank passes a certain size and then don't reduce the bet size. I suppose logic would suggest my my maximum drawdown should then be expressed as a number of bets. The logic being that if my bet size is 1/200th of my bank non reducing I can take a maximum loss of 200 units.
As for whether your 58% is acceptable, using jfc's drawdown calculator you will find that maximum drawdown increases (not surprisingly) with the number of bets you make. If you have lost 58% in two years of betting - how long do you expect to use this system? I'm guessing if you use it for 10 years you will probably experience a largest losing streak of at least twice as big as the one you've had in two years.
So the decision my friend is yours. Grasp the nettle with both hands, follow the piper into the cave and stop prevaricating. You only live twice - or so it seems.
KV

Hi Duritz!
Each way bet here. Depends????
I start counting (incrementing) the draw down when I go into the negative, that is after I have lost the bank and going on losing. At the first win the drawdown reverts to zero but the amount is kept in memory or file and if a bigger comes along it will replace it as the max DD.
I find it pointless to regard the losing runs as drawdown, after all draw_down means drawing on your own resources, and depends how much you can afford to lose before pulling the plug, not on winnings.
But each to his own here.

I find your largest drawdown strange Iomaca. When I lose the bank it's curtains. Your system sounds like the path to ruin and pressing the button at the bottom of tab home pages which says "Gambling problems". But I defend to the death your right to mean what you like when you say drawdown.

Apologies to Duritz for my use of the word prevaricating instead of procrastinating. It's the sort of mistake any idiot could make when 3 sheets to the wind and carrying on a conversation while typing.

KV

Where did I say I was losing the bank Kenny?
I was merely speaking theoretically, ie if I am running a programme of "what if"
and if losing the starting capital midway the run, that does not mean I am losing overall, the capital could have been as little $20. I have to rephrase that, the capital is always zero when I run a what if programme!!!!
Ease up man, you seem to be too tense lately!

Thanks for that guys, just used that drawdown calculator, but in what is that being expressed, are the units betting units, so if it comes up with a drawdown of 116 units, does that mean 116*whatever your bet unit is?

IE - if your bet size is flat stake $1 no matter what your bank size, does that mean the maximum DD from the peak is $116?

KV, prevaricate - no-one ever does that here!?! :)

As to procrastinating.... well I've taken the bull by the horns, am punting the sys, just at the same time trying to work out what kind of run I am in for.

Interestingly, if on the two years results of this method you bet 2% of a bank rather than 1%, the max drawdown goes to 84%, which for mine is too high.

I'll run it through flat stakes and see what the "actual" max drawdown is, in units.

Here is how to do a draw-down chart:

Get some graph paper [or get one up on your puter]. In the middle of the paper [horizontal perspective] draw a 'mean' line across the page in the middle. Mark on the top of page 12 points [dots] across it evenly spaced representing months. Name the months above the dots.

The mean line represents $0 result which might be for a starting $5000 bank for instance. Each printed graph line above or below the drawn 'mean' line represents a unit of value depending on your average bet size, the horizontal graph lines might represent $10,20.50.100 etc.

At the end of the first month if you are in say $500 profit and your horizontal graph lines represent say $20, count up 25 lines and on the vertical line that represents the first month mark with a dot. Draw a line from that dot back to the start of your 'mean' line. Keep doing that, joining the dots every month [can do one for weeks if you like]. After 12 months the lowest point reached below the 'mean' line is your max. draw-down for 12 months.

The above is for a profit and loss draw-down chart. Naturally you can do them to represent almost anything [how many pots you drink a month even].

Just change the values involved to suit your needs. It might be a SR draw-down chart for instance which will let you know SR max-drawdown [max run of outs] in 12 months. Just give everything values and away you go !!

Thanks Crash - I'd just finished doing what I said in the previous post, ie putting it in at level stakes and finding out the max drawdown. Over the 2 years and 6000 odd selections, at level stakes the max it dropped below the highest point was 79 bets.

That's a bit of a losing streak (obviously there were winners within it) - if you had a bank of X and were betting 1% of bank, non reducing, you'd obviously be down to 21% of your top amount at that point.

I think the rising/falling 1% of bank is a good idea, waddya reckon?

Although I respect your right to mean what you like by drawdown I don't think your version of it here is as worthwhile as the one (I believe) Duritz has already employed - and I'll explain why.
Your figure is certainly your worst point of the period but when testing a system and not wanting to blow the bank you need to have a handle on the worst case scenario. If you had started the system at the end of the first month when in your example you were $500 in profit your maximum loss would be the difference between $500 positive and your lowest point. When testing a system you need to know what is the worst performance from a previous highest point to a following lowest point to cater for that dreaded run of outs - which may come right at the start of the betting cycle.
I'm going to stop now because as Iomaca pointed out, all this talk of runs of outs is beginning to make me feel tense.
KV

Hi Kenny,

I was just pointing out how to do a general draw-down chart that can be applied to anything as used by stock markets and most businesses, especially large ones. They can be used for all sorts of things including any form of betting or race info. progress.

However [don't we just cringe at that word:-)], what you are describing is a theoretical model approach which can only be based on someone else's figures. How do we know they never started on the system just after an even bigger draw-down than their figures eventually show? Simple, we don't know.

Some theoretical value is there sure, but that's all. It might show max. draw-down for the last 10yrs. but the next 12 months might show a bigger one. You might start into a system at the end of a big draw-down month only to immediately fall into an even bigger one straight away.

You can't 'start a system at the end of it's first month'. When you start it is the first day surely[?].

I've avoided talking about runs of outs, but I've ended up with month's and periods. Why is punting obsessed with bodily functions?

The proof is in the punting
 

Indeed...

You're right about not knowing at what part of a system's results we enter, whether it be at the end of a bad run, middle of one, in a "normal" period (almost a completely irrelevant word, normal, when it comes to punting), or in a profitable period.

However, (there's that word again) in the absence of going back to the start of racing in Australia, and therefore the TRUE beginning of every system ever built on Australian racing, we must start somewhere. And if we start far enough back, where we entered becomes less relevant because the results since begin to stack up on their own. If, for example, I went back 10 years with this system (in a couple of months from now I'll be able to do that - without a time machine too) then it would be of little relevance at all at what point I entered, because the results since add enough info as they are.

For now, for this system, (which I started punting yesterday, 15 bets total for the day, profit of 3.3 units, pleased of course), I think a 1% increasing/reducing (weekly recalculation) of total bank is safe enough without being too wimpy - in the test period I'd have lost 58% of my bank in the worst period, but at the end of the two years ended up well in front.

Personally, I don't think two years worth of data and 6000 bets are enough. When I can, I will test this system over the last ten years, break down the results year by year, month by month, and if it wins every year, I'll start to engage the inflate-ometer on my noggin.

But, of course, as always the proof comes in the pudding, or in this case, the punting, actually outlaying the money, keeping the records, tracking the results, and - importantly - getting a "feel" for it. I am looking forward to seeing how this is going after one month, then two, see if I can get in front (have the POT figure in the Excel sheet where I'm recording the info permanently in conditionally formatted BLACK rather than RED), get the "lead" on the method and maintain it.

Good most Duritz. However [:-)], just remember these words uttered by many a pro. and punting experts all over the world. One of the biggest loss factors next to stupidity, is over-caution. Don't get too carried away with 'testing', jump in when a profit is there with reasonable expectation. It's called 'gambling'. Go for it or the profit can just as easily pass you by and turn pear-shaped!!!

Yeah that's a good point. I like profit, it's made it last two years, so I've put aside a bank for it, am punting it. After two days, 20 bets total, 2 bets in front.

OK here's a month by month breakdown of that system, in "bets" profit or loss, if you think of it as $1 on each then it's how much won or lost per month level staking this system, $1 on each, over the last two years. The months are combined, ie Jan 04 and Jan 05 totalled, etc:

Jan -39
Feb -19
Mar +107
Apr -21
May +98
June -40
July -26
Aug +31
Sep +108
Oct +2
Nov +134
Dec +83

Total = +417 over the 5800 odd bets.

I would have loved to have seen each month in profit. This hasn't happened. In fact, it's very volatile.

I'm not a big one on stats and trends etc, but given that this method is averaging about 240 bets a month and is overall showing nice profit, is it odd that it is so volatile month to month? It has won six of the months, and LOST six of the months. 50% of the months LOST.

Should I be concerned?

Good for you mate. I think you are a first [besides me] for ever admitting a loss for any month. Didn't you know there are no losers on this forum? Get with the program !!!

lol yeah we're all billionaires, no really. No-one LOSES!!

That six months losing, six months winning bit all seems pretty hit and miss to me though. I would have thought a system with a large number of bets and a healthy POT would win 75%+ months..... do you think there's cause for alarm? (said one bank robber to another)

Hi Duritz!
Did you, or can you break your stats down by weekly results?
Losing for a whole month is a bit more than I could stand but if you win two or three weeks in a month and lose the lot the rest of the time, that's a different matter.
The fact that your system is in profit, means you are doing something right, but you must find out the reason for the losing sequences, in my opinion they are far too long. Never will make it winning ALL the time. I know you said stats aren't your forte, but I'm afraid it is your only answer, and with a system, or rating like yours, it is well worth the time and effort.
Good luck

As a sadder but wiser person I would say yes.
I thought I had a great quinella system going. It averaged about 14% Profit on turnover over a period of 10 years betting on probably a third to a half of the races on offer. I watched it for two months not spending any money and it was almost embarrassingly successful. That was up to November last year. I started putting money into it in December and by early January I had lost my nerve and half my bank it lost so badly. It was a very volatile system too. I followed it up to the start of March and it lost consistently week to week so I lost interest. I actually started watching it again last weekend and it ran at 100% profit both days of the weekend and just in front today.
I should have done my stats a bit more thoroughly but I was lulled into a false sense of security by a couple of great months when I happened to start live testing.
Maybe half way between Crash's "Get into it before it falls over" and Marcus's "You must find out why it has losing sequences" there is some ground your comfortable with.

Don't know why I wrote all that - it's all waffle really. Sorry mate, you'll have to make your own mind up.

KV

Hi Duritz,

Please take my suggestions constructively.
Seems we are some of the few really putting in the research (publicly) and it's paying dividends.

Firstly, to show a profit over 5,800 bets is phenomenal.
I would be worried about the volatility, but it only depends on your personal habits. I could never endure a rollercoaster such as this, as I am in essence a confidence punter. Even a losing month is hard for me to digest financially and emotionally (discipline wise).
I recognize my short comings and have turned around my punting since that time. I also know that sooner or later I will have a couple of losing months in a row, but I can endure this if it is not a regular thing.

If you have the personal confidence and discipline to stick like glue to the system, never wavering, then I say good luck to you and you deserve every dollar profit you collect.

It's a damn fine system guaranteed.

Just curious, is this the profit over 2 years betting?

Cheers thanks KV that wasn't waffle, it's got me thinking.

Chrome - yep, two years worth (I combined the January's, Febs etc so those monthly figures are actually combined data).

Like you, I am not one of those firm, like glue types who'll resolutely stick with it through thick and thin - I tend to waver too. I really think I'll need this to make a constant profit or I might not stay the course.

It's making me think hard about what to do - what I'd certainly like to do is test it over longer.

Going to sleep on it.

Duritz.

[QUOTE=Duritz]
Jan -39
Feb -19
Mar +107
Apr -21
May +98
June -40
July -26
Aug +31
Sep +108
Oct +2
Nov +134
Dec +83

Total = +417 over the 5800 odd bets.

Should I be concerned?[/QUOTE

No if it's not the rent money your using as capital to punt with, yes if this bird flu pandemic goes human to human [I'm stocking up the food supplies and candles :-).]

What I find interesting in your figures is the loss/skinny profit periods. It certainly reflects my bad periods and good periods to a large degree.
OK we have Jan. And Feb. Hot as hell and lots of 2 yr. olds running around as 3yr. old with ************ all form. Difficult period. March, well those 3 yr. olds have some form now and there is good prize money around [less dodgy races and trainers doing their thing]. More money around basically and more reliable form.

Winter, the flip side of summer with very little money around and then an upswing into the money period of spring. Xmas period into Jan. early Feb. is probably a good time to do something other than punting. I certainly wind down my punting $ outlay during that time. A tough period.

I find all that timing reflects my system and/or handicapping efforts similarly and that in itself is interesting. Well maybe it is if many other punters have a similar experience with different times of the year.

Yeah I tend to concur with that - also in those months like Jan and Feb, in the open grades the fields are thin (have a look at some of the fields on a Wed at Sandown) and the favourites short, that makes it hard to make a buck.

Perhaps further proving that - my handicapping method has its highest strike rate of winners on top & winners in top 4 in Dec, Jan, Feb - yet the sys can't win in Jan Feb, and its derived from the handicapping method. This says to me that in those months the winners are easily sorted, and go around at their right odds.

A Worrying Sign
 

OK this is not only worrying for me, guys, it's also worrying for everyone else whose been in on this thread - see that list I published previously, the month by month breakdown? Well, that's only five of the months as losing months, not six. I don't know how I counted that as six, and I don't know why no-one picked up on it!!!

It must be because I'm so highly respected here, held in such high regard and esteem that as soon as I said "six months losing" no-one thought to question, because surely this genius, this Goliath of an intellect could not have made such an elementary, moronic error as to miscount the plusses and minusses!! :o

Well, it seems I have.

So, five months losing, seven months winning. That's becoming a little more easy to handle.

To break them down, here they are:

Losing months: -39, -19, -21, -40, -26
Winning months: +107, +98, +31, +108, +2, +134

I'm a lot happier right now, looking at that - because also there might have been losing months there but they're not that big. Even 40 bets (though 40% of your bank if betting flat stake) is not so bad when you've got an increasing/reducing bank. It still means this system is going to be volatile, but some of those winning months are really great winning months.

So, anyway, here it is broken down month by month for each of the two years, rather than both combined:

2004

Jan -14
Feb +2
Mar +64
Apr -16
May +26
June -36
July -57
Aug +16
Sep +93
Oct +22
Nov -1
Dec +45
Total = +144 for the year

2005

Jan -25
Feb -21
Mar +43
Apr -5
May +72
June -4
July +31
Aug +15
Sep +15
Oct -21
Nov +135
Dec +37
Total = +273 for the year

So, both years in the positive, healthily so, esp 2005.

Losing months, broken down

2004: -14, -16, -36, -57, -1
2005: -25, -21, -5, -4, -21

So, of the 10 losing months, 3 of them are virtual chop-outs (and with better than tote odds, surely would be chop-outs). That means there's actually only seven of the 24 months total which are REAL losing months, if you will, or about 30% of them. I said earlier I wanted 75% of the months in profit, well we're getting closer now.

The seven losing months go -14, -16, -21, -21, -25, -36, -57. Those two worst months were June and July of 2004 - the worst run of the method, quite clearly. Loss of 93 units over those two months. I guess what that shows is that you need to minimize damage in times of loss.

To illustrate this, I've compiled a little example. If you'd started with a $1000 bank and betting 1% reducing/increasing ($10 start point and a $10 minimum no matter what the bank) on each and started at the start of '04, by June 1 your bank would be $1630 (profit $630), and your bet size $16, when the losing streak begins. Beacause of reducing/increasing bet size, by the end of July when you've lost 93 units over the two months, your bank is down to $690. If you had a "non-decreasing" clause in there, your bet size would have peaked and held at $21 from early April, and on July 25th you'd have lost the last of your bank. So, increasing/reducing has kept you in the game and only lost 31% of your initial bank, non-reducing has wiped you out.

Interestingly, if you'd just level staked $10 without increasing/reducing at all, at the end of July your bank would be $690. So, not much difference if you just flat stake but then you don't give yourself a chance for big profits as you increase. For me, I think this shows that the 1% increasing/reducing has the scope to take advantage of "good times" and the ability to cope with the bad runs without a wipeout. However, you need the minimum bet rule. I set it at $10 for the sake of simplicity of maths for this, it of course depends on yourself but realistically you wouldn't want it below $5. From $5 to professional punting levels is like from the Manangatang picnic gallops to the Cox Plate. As I said, I've used $1000/$10 as a good point for maths.


OK, just re-read all of that. I don't even really know what point I am trying to make, however I think what this has done is let me learn a lot more about what I am potentially in for with this system. This is not going to be a steadily increasing bank like the graphs on the glossy pages of the system sellers, which is why I think th 1% increasing/redecing is perfect for it. It runs with the peaks and troughs and adjusts itself for them.

Waht do you guys think of those figures?

http://home.iprimus.com.au/jfc2000/drawdown.htm

Reminding or introducing any who might be interested about this toy.

If you want to try out the situation in the Triple your Bank thread the relevant parameters are:

20 = % - Chance of Winning
7 = Decimal Odds


One thing worth noting is that Maximum Drawdown is never less than Longest Run of Outs. That makes the Drawdown far more useful.

For example a few fluke wins could break long run of outs and give false assurance about a dud system. Whereas the Drawdown is more immune to flukes.

What do we think of the figures?

All over the shop, but if the system is making a profit, great!
The nitty-gritty for you though is: I bet this system is not your only bets [no exotics or other bets etc.?] and are you keeping a record of those? :-))

From a true statistical perspective would these results really be deemed volatile?

I think it is a bit like the stock market in as much as the degree of (perceived) volatility will vary with the time span being examined. If you looked at these sort of results in three monthly periods after (say) ten years then it would possibly appear a very stable set of results?

As Chrome alluded to, I think the hardest thing for punters to come to grips with is that profit making doesn't necessarily follow a nice straight line - which is why you see some punters changing systems/methodologies as often as they change Irish underpants.