2012 : Breaking With Tradition (Just a Little) - Part I

MAFL usually changes a lot from one year to the next, as Funds are reviewed, reviled, revered, retired, revamped and relaunched. Maybe this year it's time for more stability in the roster so, like the Shadow tipping algorithm, I've decided to show a bit of loyalty and retaining both of last year's Funds albeit with a few tweaks to each. I'm also going to trial an experimental Margin Fund, details of which will be in a subsequent blog. If you're a MAFL Investor with carryover Funds now's the time to decide whether MAFL 2012 will remain a part of your balanced portfolio of investments. (Of course if the Mayan Prophecy believers are right, none of us will get the chance to reflect on the wisdom of our investment decisions in 2012.)

The Line Fund Stays, But Will Level-Stake Commencing in Round 5

Last year the Line Fund returned a small profit, so retaining it is surely among the least contentious of the MAFL decisions I've made for the upcoming season. I am making a few changes to it, however. As noted in last year's end of season blog, the Line Fund of 2011 suffered from early-season wagering, so I'm imposing a 4 round moratorium on Line Fund wagering this season - in other words, the Line Fund won't start betting until Round 5.

Kelly-staking the Line Fund was also suboptimal in 2011 - as it has been for most seasons historically - so the other change for this Fund for 2012 will be that it level-stakes at 5% of the Fund for every wager it makes.

Had both of these restrictions applied in seasons 2007 to 2011, this is what the Line Fund performance would have been:

Absent a level-staking approach, remarkably the 2008 result would have been a loss, notwithstanding the 57.7% success rate. The 2009 result for the Fund is not ideal, but the loss in that year is overwhelmed by the profits in the preceding and succeeding years.

If you want to convert the ROIs shown here to RONFs (ie actual profit) assuming 5% level staking then multiply the ROI provided by the number of wagers by 5%. So, for example, the 2011 result implies a RONF of 4.1% x 73 x 5% or about 15%. So, a $1 investment in such a Line Fund would have returned $1.15 at the end of the season. The actual Line Fund of 2011 returned about $1.09.

Note also that the Line Fund will, as in years past, wager only on Home teams. 

If history's a reasonable guide, the Fund should wager on between 45% and 60% of games from Round 5 onwards which, given that we'll have 9 games a round for most weeks in 2012, means we should expect about 4-5 wagers per week.

Currently I'm thinking that this Fund will carry a 50% weighting in Investor Portfolios.

The Head to Head Fund Stays Too, But Wagers From Round 6 and Now Ignores Short-Priced Favourites

Retaining last season's Head-to-Head Fund was a much more difficult decision. It too suffered from early-season wagering, but also from the pricing cap and the probability adjustments I imposed on the Fund algorithm, these restrictions collectively allowing the Fund to bet early when it didn't know enough about the teams and preventing the Fund from betting later in the season when it had identified what turned out to be genuine pricing anomalies.

So, firstly, I'm imposing a 5 round moratorium on wagering for the Head to Head Fund this year, I'll make no adjustments to the Fund algorithm's probability estimates, no matter how different they are from the bookmaker's, and I'm also removing the $5.25 ceiling on the acceptable price for potential wagers. In fact, based on a reanalysis of the Fund's historical performance, I've decided to replace the ceiling with a floor (not, I admit, a sentence you'd want to hear from the $1,000 an hour architecture you've commissioned to design your new house). The 2012 Head to Head Fund won't bet on a Home team - and Home teams are still all it will wager on - unless it's priced at $1.50 or more.

Historically, wagers on such short-priced home team favourites would have represented a significant proportion of the Fund's wagers - between about 50% and 75% by value of the wagers in individual seasons from 2007 to 2011. Collectively, these wagers would have produced losses in 2007, 2008 and 2011, a small profit in 2009, and a healthy profit in 2010. Across all 5 seasons, wagers on such teams would have represented about 66% of all wagers by value, 40% by number, and would have generated a positive ROI of just 0.35%, thanks almost entirely to the bonanza from 2009. That's a tiny and uncertain reward for a very large outlay.

The new Head to Head Fund would have recorded the following historical performance:

That column of black ink under the heading ROI is nice, but it's important to recognise that each season's results are from a relatively small number of wagers. The 2010 results are for just 33 wagers and those for 2009, he year in which the Fund would have been most active, are for just 54 wagers. You wouldn' t need to flip many results to flip an ROI from black to red.

In particular, the positive result for 2011 is due to a handful of big wins, a fact that is demonstrated by the second-last column of the table, which shows the percentage of all winnings that are represented by the season's 5 biggest wins. For 2011 the figure is 80%, meaning that the 5 biggest wins of the season represent a significant proportion of the returns from winning bets for the season; the remaining 10 successful bets, combined, contributed just 20%. The last column of the table shows that the losses in 2011 were more evenly spread across different wagers since the 5 largest losses represent only 40% of all losses from unsuccessful wagers. This fact only adds to the precipitous nature of that 11% ROI for 2011.

The Head to Head Fund will wager only on Home teams and will Kelly-stake in 2012, as this staking approach has historically outperformed level-staking for this algorithm. I'll once again be using a divisor of 5 for the Fund, which should keep the average bet size in the 3-4% range that you see in the table.

(A quick refresher on Kelly-staking: the full Kelly-stake, as a proportion of available funds, for a wager priced at $f on a team whose victory probability you've assessed as p is (pf-1)/(f-1). This formula can produce startlingly large fractions. For example, if you can secure $3 for a team you've assessed as a 50% chance of winning - meaning that you believe a fair price is $2 - the recommended Kelly-stake is one-sixth of your available funds. A more cardiac-friendly version of Kelly-staking is fractional rather than full in which the recommended wager is divided by a constant; I'm using 5 for this purpose.)

This small average bet size, combined with the Fund's relative inactivity - it's likely to make only 2 or 3 bets per round once it starts wagering - means that the turn for the Head to Head Fund should be low in 2012, somewhere in the 1.1 to 1.7 range. This is far smaller than the turn of 6.9 recorded by the Head to Head Fund last year, and will mean that any negative ROI for the Fund will convert to a much smaller loss in RONF (ie real) terms. It also means, of course, that any positive ROI will not enjoy as large a multiplier. 

As a further buffer against underperformance by the new Head to Head Fund I'm proposing that this Fund initially carry only a 30% weighting in Investor Portfolios.

What About the Remainder?

The remaining 20% of Portfolios will be invested in a new experimental Fund, called the Margin Fund, details of which I'll provide in the next blog.