Tuesday 6 November 2012

Documenting the arbitrage

The ongoing price difference between InTrade and BetFair offends me. So I decided to do well while doing good. InTrade verified my paperwork. I sent them some money. The Bank of New Zealand delayed things because sending $3k to one 'internet gambling' site followed by an attempt at $1k to another [their term; I say they're futures markets] was deemed suspicious - and rightly so. But, it meant that I had to wait until their call centre opened up at 6:30 am to start playing.

Once you make the plays and figure out the charges, it's fairly obvious why price differences between these two markets can persist. Recall that it isn't easy to set up the accounts in the first place.

I sent AUD$1000 to BetFair. They took a 1.5% credit card fee out of that. I put $1000 on Mitt Romney to win at 4.4 [I had a standing account balance]. If I win at BetFair, they charge a commission on winnings. Since I do not trade frequently, I get no discount on that. So if Romney wins, BetFair will take 5% of my winnings. This adds up on an arbitrage play.

I sent USD$3020 to InTrade. They took $20 to clear a negative balance that was sitting there. InTrade charges a $4.99 monthly fee to keep an account and recently changed the rules so that dormant accounts of zero balance are not charged fees; it was simpler for me to pay the $20 than to set up a new account. I'll just have to remember to withdraw everything from my InTrade account when the play is completed so that it doesn't just erode back down to zero. But I'm going to count that $20 as part of the cost of the overall play. InTrade charges zero fees other than the monthly trading fee. Fortunately, InTrade underprices the candidate who is likely to win. I bought 448 shares of Obama at $6.69 per share.

The tally thus far [updated with a couple of corrections]:

Position$1,000 AUD + $2,997 USD$4,034 USD
Account Fees$15 AUD + $20 USD$36 USD
Cost:$4,070 USD
If Obama wins (gross)$4,480 USD
If Obama wins (net of costs above):$410 USD
If Romney wins (gross):$4,400.00 AUD
Less 5% charged on winnings:$4,230.00 AUD
Net earnings if Romney wins:$318 USD

I'm unfairly charging the $20 fee against InTrade. But once you put in the 1.5% credit card fee and the 5% tax on winnings, it's not hard to justify a reasonable price spread across the two markets. So my earlier tweet was wrong - in particular, it missed the 5% winnings tax at BetFair. If Obama wins, I'll do well. If Romney wins, I won't lose anything except my time.

So why does the price spread across markets persist?
  1. It's a hassle to set up accounts
  2. BetFair's tax on winnings limits arbitrage opportunities. 
There's still a bit of free money sitting in the price gap between markets, but I don't really see it being worth putting more money in given BetFair's fee structure. 

I have a fair bit of cash balance sitting at iPredict; I'm placing a large order long Romney and short Obama there - iPredict is currently saying a 77% chance of an Obama victory. They charge trading fees capped at $5 per month per trader, a 1.75% fee on credit card deposits (irrelevant for me as I have cash sitting there), and a withdrawal fee of 2% if you have positive earnings (sunk for me, but relevant for you if you don't have an account there). 

I've standing orders at iPredict selling Obama at $0.76 and buying Romney at $0.24. If anybody takes those up, I'll send more money to InTrade to lay it off. But this involves the hassle of keeping an eye on the iPredict order book so I can move quickly if something happens in the election that moves the odds.

Arbitrage isn't supposed to be easy....

Note: I do have some unhedged currency risk: if the USD/AUD exchange rate moves substantially against me, or if the NZ dollar appreciates substantially against either, I could in some worlds lose a bit. Suppose that a surprise Romney win makes the Australian dollar tank against the US dollar. My winnings in Oz, converted to US dollars, won't look so nice.

Update: I switched to offloading Romney risk onto iPredict, New Zealand's election stock market. It's thinner than BetFair, but if you put up a big order, it might be picked up. And iPredict's tax on winnings is lower than BetFair's - 2% withdrawal fee if you have positive earnings.

Current position then is:

  • +2114 Romney at iPredict at $0.24 ($NZD, mix of long Romney and short Obama)
  • +670 Obama at InTrade at $6.71
  • $1000 on Romney at 4.4 at BetFair.
If I haven't screwed up the spreadsheet, and net of all expected fees, I'm up about $120 USD if Romney wins and up $710 USD if Obama wins. Total stakes USD$5950. So a short period return of 12% if Obama wins, 2% if Romney wins. I expect Obama to win, so I don't mind that exposure. 

Remaining risks not here accounted for: 
  • Currency exchange risk: If the US dollar tanks on an Obama win (exceedingly unlikely), I'll take a hit; if the Australian dollar tanks on a Romney win (exceedingly unlikely), I'll also take a hit. Or, if exchange rates just happen to move substantially between now and settlement. The $NZ is relatively high now though - it would be surprising if it spiked upwards.
  • If the bank's spread on exchange rates is wide enough, 2% gains on a Romney win turn quickly to zero percent. 
  • iPredict isn't a perfect hedge: the other markets pay off on Obama and Romney; iPredict pays out based on which party wins. 


  1. hi, I'm a regular trader on Intrade, and also very puzzled as to the gap between Intrade and the British bookies. Unfortunately living in America I can't open an account on Betfair/Pinnacle, so I can't arbitrage that gap.

    Just wondering, instead of buying Obama and Romney, why didn't you just buy Romney at 22% on Betfair and sold Romney at 35% on Intrade? wouldn't that guarantee almost 50% profit?

  2. I had thought that InTrade didn't allow naked shorts. If shorts are covered, there's no difference between being long $0.25 and short $0.75. If InTrade doesn't require that shorts be covered and instead uses margin calls, then I've screwed up. But how could they enforce a margin call?

  3. well, that's true, shorts are covered, so in my scenario, in order for you to sell R at 35, you'd have to have 65 in your account locked in as margin requirements. But still, considering the actual trade, you'd really buy for 22 and sell for 35, meaning a 50% profit, even though you'd have to have much more money in your Intrade account for margin requirement. Does Betfair have limits on individual bets or deposits?

  4. I had a look at this yesterday - not doing it for reasons which include a 5.5% round trip on $ from GBP from my bank.

    I don't think betfair comm is that bad (although I pay <3% at betfair). 5% comm would move your Romney probability (@4.4) from 22.72% to 23.64%.

    On both sites you have to put up funds to cover your liability, so your % profit should be measured against this, so buying at 22, sell at 35 and putting up 22+65 => 13/87 (15%) return.

    Eric bought Obama @ 66.9 and (implicitly) sold at (1-1/4.4)=77.28, making => return 10.38/(66.9+22.72=89.62) (11.6%)

    So that would have been a better trade, but because of the wider price difference, not because of the relative difference in prices. Also, if obama was bought at 66.9, I think you would only be able to sell Romney at under 33.1, not 35