Using Quant to predict the World Cup

(click to enlarge)


You might as well applaud such a move as nothing much will get done during the World cup month. Possibly meaningless to Americans, but hey, they call the stupid baseball thing World Series and only American teams are playing???!!!

http://www.hotcute.cn/wp-content/gallery/Chinese%20sexy%20supermodel%20Cica%20Zhou%20Weitong/Chinese%20sexy%20supermodel%20Cica%20Zhou%20Weitong%20(13).jpg


Whilst this report should be taken with a pinch of salt, we find it an interesting exercise and an ideal opportunity to lightheartedly explain Quantitative techniques and demystify the typical

Quant framework.

I am so fed up supporting England during World Cup tournaments, so I am deserting them and going for the very well balanced Spain side, as well as cheering on all the underdogs: Ghana, Japan, South Korea, North Korea, ... to name a few.

I was having lunch with an Englishman and a Brazilian yesterday (yea, business meeting in HK) and they were giving me a hard time when I said I will be supporting Spain. Geez ... I said that if asians were only "allowed" to support their actual country team, then we might as well forget about supporting any team for the World Cup. I said its lucky for them to be accidentally born in England or Brazil. I then asked them if their country teams did not make it to the World Cup for 10 or 20 years, what then, who would you support? That shut them up.

You will note that the headline of the JP Morgan report has England as the likely winner even though on team's strength analysis Brazil is tops and in betting Spain is favourite. When arguing over World Cup teams, its highly emotive and hence you have to use data only to remain impartial. Quantitative methods will use past data and then project ahead.

Their goal is indeed to highlight potential World Cup winners by applying quantitative or mathematical methodology traditionally used with balance-sheet, valuations and consensus information to data from the football world. To do so, they focused on data including:
• probabilities to win from a range of bookmakers and exchanges
• official FIFA World Rankings
• results from previous World Cup tournaments and qualifying competitions

http://www.hotcute.cn/wp-content/gallery/Chinese%20sexy%20supermodel%20Cica%20Zhou%20Weitong/Chinese%20sexy%20supermodel%20Cica%20Zhou%20Weitong%20(15).jpg

In practice, Quants tend to use 4 types of information in their mathematical models:
1. Valuation metrics
2. Market and Analyst sentiment
3. Company fundamentals
4. Price trends

J.P. Morgan Cazenove "Normal' Quant stock-picking Model
VALUATION METRICS MARKET
- PE vs the market
- PE vs the sector
- Forecast growth
ANALYST SENTIMENT
- Recent change in analyst sentiment
- Recent change in analyst growth expectations
- Recent change in analyst recommendations
COMPANY FUNDAMENTALS
- ROE
- Company Risk
PRICE TREND
- Long term trend
- Short term trend

Source: J.P. Morgan

They then translate the above Model into a football-specific Model.
J.P. Morgan Cazenove Quant world cup-picking Model
"VALUATION" METRICS "MARKET
- "Market" Valuations
- FIFA World Ranking
& ANALYST" SENTIMENT
- Result Expectations
- Recent Team Shape
"COMPANY FUNDAMENTALS"
- Consistency in Market Sentiment
- J.P. Morgan Success Ratio Indicator
PRICE TREND
- Trend in probability to win
- Trend in FIFA's Ranking

Source: J.P. Morgan

If you can get the report, its a lot of fun. They looked at the actual FIFA world ranking and then looked at the usual probability in winning. Regressed them and somehow, some of the teams' rankings do not match up with their win probabilities. Portugal, Netherlands and Greece offer a disagreement with high FIFA World Ranking and low indicated probability to win the World Cup.

England, Argentina and Ivory Coast also offer disagreement with a low World Ranking and an indicated high probability of winning. According to the FIFA World Ranking Factor, Brazil, Spain Netherlands and Portugal are most likely to win the World Cup.

http://www.hotcute.cn/wp-content/gallery/Chinese%20sexy%20supermodel%20Cica%20Zhou%20Weitong/Chinese%20sexy%20supermodel%20Cica%20Zhou%20Weitong%20(188).jpg

Hence based on that alone, a betting person should favour England, Argentina and Ivory Coast, ceteris paribus, and go against Portugal, Netherlands and Greece. On a side note, Greece might play luar kulit (out of their skins) owing to the homeland crisis and on the pretext that they may be booted out of EU soon (hence no more invite to Euro championships???!! ; ) )

They then look at bookmakers' odds to ascertain value against what the computer predict as their real value. They then plat charts based on the 6 month trending winning probability and a 3 month trending winning probability - a lot like 30 day MA and 60 day MA. Hey, a trend is your friend, it applies in everything.

According to the Trend in Probability to Win Factor: Slovenia, France, Ivory Coast and Greece are the most attractive options, having received the greatest increase in probability over the past 6 months.

According to the Trend in FIFA’s Ranking, Algeria, Slovenia, Serbia and Slovakia have the biggest change in World Ranking Points and should be preferred.

But of course, as wonderful the data may be, it still depends on the weightage you assign to each of the 4 factors. Herein lies the problem, JP Morgan assigned equal weightage to all 4.

The very funny part is that the last 2 pages uses the quant models to predict every match. In the quarter finals these are teams, according to the quants:

Netherlands vs Brazil (Netherlands will win)
France vs England (England will win)
Argentina vs Slovenia (Slovenia wins in an upset)
Italy vs Spain (Spain wins)

[zhouweitong-35.jpg]

In the Semis:
England will meet and beat Holland, while Spain will edge out Slovenia.

Ta-dah, England beat the crap out of Spain in the finals. Yea, right!!!??

No comments:

Post a Comment