Monday, 10 September 2012

Slave To Records

While there are problems with its application in the world of sports betting, the Kelly criterion is proven and widely accepted as the optimal staking strategy. It is comprised of two factors - winning probability and win/loss ratio.

When it comes to sports trading, the strategy of using Half (stakes) Kelly or even Quarter Kelly is often suggested. Returns can be volatile, but this volatility is reduced by using a fractional form of Kelly. For those unsure of their edge, staking close to 2.5% of your bank is the usual advice.

The phrase 'Calendar Kelly' came into being to describe a situation where an investor deviates from his chosen stakes, not for any technical reason, but because they are influenced by some ephemeral or psychological reason.

The problem with allowing emotions into your trading decisions, is that you no longer are playing the markets optimally. One of the assumptions for Kelly (or a version of it) is that the investor is able to maintain historical performance, but this is of course not the case if you chop and change your strategy every few weeks.

Note that Calendar Kelly does not refer to the quite logical decision to reduce stakes towards the end of a season, or in situations where a contest means more to one team than another. Such situations mean that your ability to accurately estimate one of the key factors (win probability) is diminished, and it is quite reasonable to reduce your stakes. There is a logical reason for doing this, but applying Calendar Kelly in its true sense is not logical. It means that you are not staking optimally.

As most readers will know, this topic was triggered by Mark Iverson's admission that in most months, he reduces his stakes in the last seven days. On July 25th, he Tweeted: 
"Feel like I'm reading this match well so far. No need to push it coming up to the end of the month mind", 
a statement that I felt should not go unchallenged, given the influence that Mark has in the sports trading world. In the same way that religious people slip mentions of their god casually into conversations as if their god was a fact, I felt that Mark's implication that it somehow makes sense to ease up based on the calendar was poor advice, and that with 1,000+ Twitter followers, the comment should be challenged.

After a few exchanges in which Mark tried, but ultimately failed, to justify any logic in his 'strategy', you can read it all here, Mark finally admitted that this course of action is emotionally driven, which of course, we all knew all along.

So there is no debate that from a purely financial point of view, Calendar Kelly makes no sense, and leads to lost profits. The argument for the defence appears to be based on the fact that Mark's confidence is so fragile, that a losing calendar month would be devastating. Yet this is a professional trader, who, I believe, has gone six years without a losing month. (It would be interesting to know how much it has cost Mark in the long run to preserve that record). I am actually more open to understanding this approach from the casual / part-time trader than I am from a professional. I think most of us would soon be in trouble if we took this approach at our place of work. For those of us with bosses, we would probably soon be seeking alternative employment, and for the self-employed, then clearly they are costing themselves money.

Mark wrote a post yesterday which you can read in full here, and addresses the issue of easing up for 84 days of the year. He wrote:
Well the truth is that even by taking my foot off the pedal for 40% of the last week in every month, I’d actually only be losing 34 full days. When you consider that most people in a full time job have 2 full days off a week it doesn’t seem to me that I’m losing out much at all.
How I wish I had a full-time job with two full days off a week! That aside, and without getting too much into why most of us are better served by trading part-time, if you choose to trade full-time, you are choosing to work an unconventional schedule, and as with any 'professional' activity, you need to tune that schedule to optimise your productivity, in the case of trading - your income.

If you are an expert on World Cup Finals, it would be the height of folly to book a holiday from mid-June to mid-July 2014. If your expertise is the English Premier League, then schedule yourself to work most Saturdays and Sundays during the season, with midweek sessions as needed. My point is that part of being professional is taking the time to schedule yourself to produce the best results. Obviously no one can trade 24 hours a day, seven days a week - at least not optimally, but the suggestion from Mark that allowing the calendar to dictate to you when you take your easy days is sometimes acceptable is nonsense.

Mark says:
 "I can’t control when the events I want to trade are scheduled, I have to manage my life around them and it just so happens that 2 of the tasks that need to be crammed in are taking a break and doing some data management"
What Mark, and all of us CAN control, is what events we put on OUR schedule. Not many sports are played with a non-overlapping schedule, so we are already in the position of often choosing which games to trade. For example, as I write this, there are nine NFL games in progress. So what we can and should do, is schedule our breaks or our 'data management' days for when it is best for us.

Clearly someone as organised as Mark doesn't wake up on the 25th, and suddenly realise he needs to cram in seven rest or 'data management' days, so can the decision (and the financial cost) of easing up be justified by the 'benefit' of recording a green month every time? Mark writes: 
that if I suffered multiple losses towards the end of a month that the emotional side of this game would rollover to the next.
Really? Do multiple losses towards the end of a week rollover to the next week? Does a losing day rollover to the next day? What's so special about a losing month? It's somewhat surprising that a professional trader would suffer from this 'on-tilt' problem at any time, as it's one of the first things you need to learn how to handle in trading.

Mark has his apostles though. James, who rather amusingly uses the Twitter handle @CricketTrader (yet needs to attend Mark's cricket trading course, while he would be better served attending a 'learn how to write and spell' course) writes:
"Casssini come across like a 12 year old schoolgirl". 
Having never been a 12 year old schoolgirl, I have no idea what James is talking about, but clearly it's a world that James is familiar with.

The more masculine and mature John wrote a very long comment, which I repeat here:
Hi Cassini, I’ve been a reader of your blog for many years and have come to respect you views and opinions on many things. Pointing out the wrongs of sites such as fgbna and the forum post highlighted is what keeps my coming back, but I have to say you’ve started to come across a bit big headed.
As far as I’m aware you have never used trading as your main form of income. The rights and wrongs of doing this are a topic for another discussion but my reason for pointing this out is that until you have, I don’t think it’s fair for you to tell others what’s right and wrong about the way they trade full time.

I traded Betfair for a number of years as a hobby until one day I lost my job. While looking for employment I upped my trading and found I could make more than I thought – I ended up taking a part time job (2 days a week) to cover bills and topping up the income from Betfair to match my full time wage. I used my spare time to study another qualification and can honestly say the 2 years I carried this out for were the best of my life (so far!). I’m now back in full time employment and trade Betfair as a ‘second income’ rather than a ‘hobby’ – in other words I take it a bit more seriously than I did before.
Some of the lessons I learnt during these 2 years were invaluable. I would never have understood them from textbooks or reading blogs/articles on the internet. It’s only from being put in such a situation that you can fully understand and appreciate what it’s like to trade for a living (or for the majority of your income).

I’m sure the majority of your readers (as you point out, they are all intelligent chaps after all) understand there is no logic to trading differently at the end of the month. I’m sure even Mark understands this himself. But for him it works and when your income depends on how you trade, funnily enough, you carry on doing things that ‘work’ irrespective of how logical they are. One of the biggest lessons I learnt was it’s very hard to change the way you trade if what you do is bringing you a profit. There were parts of my trading I could see would benefit from tweaks here and there but to do so would mean experimenting – something I had neither the time or bank roll to do. (Arguably the issue here is I should not have been relying on trading as an income if I didn’t have the bank sufficient enough to experiment with, but again a topic for another discussion and it was a situation I ended up in rather than choosing due to losing my job). When you are faced with a)trading as you have done which is making you an income or b) tweaking it which may increase profits but also runs the risk of failing, the option most people will chose when their meal ticket is on the line is a). If Mark has found that for psychological reasons he likes to reduce his trading at the end of a month then the maths don’t matter (even thought they should). If he is bringing in profits from working this way then he is probably loathed to change his approach when the downside would mean loss of confidence and other knock on effects that will then harm him (and his family) financially. Interestingly, now I’m back in full time employment and the trading income is being used to increase a bank roll rather than pay for clothes, I have been able to experiment with those tweaks. Some worked, some didn’t. Overall I have almost doubled my income through trading since being back in full time work. However, when I relied on the income I couldn’t have afforded some of the losses I made in this tweaking phase. It would have crippled me both mentally and financially. I would have probably given up trading and gone back to full time work and this comment would have never existed! As you haven’t relied on trading as an income I’m not sure if you will get my point.

So rather than berate people like Mark, why not try and understand his reasons for doing it? Creating a discussion on the topic of why you would do this and maybe how you could avoid such illogical behaviour to improve your trading would be a far better read than the now tiresome month end jokes.
A little less of the ‘Cassini knows best’ attitude and a little more humbleness would be a good thing. After all, none of us should be foolish enough to think we know best. Hope you get my point and don’t take it personally! I read your blog because you challenge others so keep that up – just maybe do it a bit more constructively.
A couple of comments in reply.
"I have to say you’ve started to come across a bit big headed"
This is a bit disappointing. My intention from the start has been to come across as 'a bit big-headed', so for this to only be apparent recently is a little upsetting.  
As far as I’m aware you have never used trading as your main form of income... until you have, I don’t think it’s fair for you to tell others what’s right and wrong about the way they trade full time.
A fair point if I were trying to tell anyone how to trade, but I am not. I am simply pointing out that there is another side to the story and that implying to 1,000_ followers that it is acceptable to trade less than optimally for one week every month is not necessarily good advice.

Also, I have never berated Mark. As a part-timer, as John correctly points out, it would be a bit like your average Sunday cricket player berating Muttiah Muralitharan on his bowling action.

As for less of the "Cassini knows best" and more humility suggestion, is that really what my fans, I mean 'followers' want? Do people flock to this blog to read bland thoughts or do they gather here in their multitudes to hear strong opinions and a little self-confidence? I jest. I know very well that I'm not perfect - I am way too modest for a start ;)

John writes some good stuff about his full-time trading experience, especially the 'too busy to experiment' part. How many of us are too busy earning a living to look for a better job? Congratulations on doubling your trading returns since the pressure was off. Another argument in support of my opinion that trading as a second job is the way to go. 

Interesting too, that John mentions bank size - insufficient to allow experimentation, because of a fear of losing a large percentage of that bank which, when you have no regular income, is not easily topped up. While I have never had to top up my bank, and probably keep too much in my accounts, I have perhaps underestimated how key protecting the bank is, if trading is your only source of income. The fear of losing it will vary by individual. Someone with liquid assets of  £1 million is going to have a different mindset to someone with more limited resources.

Anyway, to wrap up this whole issue, perhaps it is fair to say that, as John and others have suggested, if you are trading for a second income or as a hobby, then you have the freedom to apply the same strategy every trading day. The occasional disaster doesn't matter too much in the whole scheme of things, but if you are full-time, the pressures that a (self-proclaimed) disaster would bring mean that trading less than optimally is a price worth paying. Safety first. From my tennis playing days, I know it's often a lot easier to play a great shot when nothing is at stake versus in a meaningful game.

Mark's approach works for Mark, but I still consider that being influenced by arbitrary short-term periods is poor practice that most traders would be best served avoiding. Other successful (and full-time) traders I know keep detailed records, but they are not slaves to them. This is the key call-out here. If a thirty day period happens to net out to a loss, then so be it. Who cares? They are more interested in totals at the end of a season than for calendar months, and I believe this is the best approach provided your staking is appropriate given the size of your bank and reserves. 


AL said...

i still don't get it.

if you are using a kelly staking and you start losing, your bankroll is dropping by what 2.5% each time? you wont be wiped out as your expected profits should more than compensate you!!!

in July i had only 2 days which garnished most of my months profits, the rest were either break even or losing days.

John said...

Sorry, the ‘big headed’ comment was meant to be somewhat tongue-in-cheek and I had put a smiley face at the end to try and emphasise this, but Blogger seems to have changed it to a square!
That’s one of the problems with the internet and blogging/forums. Sometimes the context in which something is said is lost – especially when trying to fit it into 140 characters.