A few weeks ago a contestant raised what they considered to be a serious error in the reports. They did not say (but probably thought) that if something like this is in the weekly output report, does it mean that we should be skeptical of the announced winners at the end of the contest?
I think in this case, it’s a matter of a misunderstanding of what the report was saying, rather than an error in the spreadsheet. (And I think the accuracy of the final results will be pretty good.)
To be specific, a contestant noticed that the XYZ stock company had finished the last week at $10 / share, and the end of the current week it was $11. The percentage change of XYZ for the week is pretty easy to calculate: (11 - 10) / 10 = 10%, which is how any stock table in a newspaper would record percentage change in any value. A 10% gain in not to be sneezed at, but is unlikely to garner attention. But the report actually calculated that it was a top performer for the week with a reported 20% change, enough to land it in the top 5 discussion. How could this possibly be right? If you could look at another stock, the ABC company who went from $1 to $1.1, which is an equally big percentage change of (1.1 - 1)/1 of 10%. But unless they asked about ABC specifically, they are unlikely to see what the percentage calculation used by the report would say. In this case the report writer figures that ABC is a 2% increase, and almost certainly not going to make the top five winners for the week and be reported on.
What’s going on is that the report % change is always the change in a stock as a percentage of THE ORIGINAL PRICE at the start of the contest. It turns out that both ABC and XYZ happened to start at $5 a share and had a very divergent history.
Just why this method is used now was discussed at some length in the middle of last year’s contest. We had been using the “newspaper” percentage change calculation to rate individual stock performance. Week after week the “winners” list selected stocks that had lost a lot of value, in one case only a few tenths of a percent of their original price. This stock could often have a huge newspaper percentage change, but they couldn’t possibly make that much of a difference to our contest, as they amounted to a small amount of actual cash.
What matters in how much money a particular price change made (or lost) for the portfolios. That is NSHARES * (THISWEEK - LASTWEEK), which varies depending on how many shares a portfolio bought in the original allocation at the start of the contest. If you want to make that a portfolio independent percentage, replace NSHARES with ALLOCATIONFRACTION / ORIGINAL PRICE, and take out the portfolio dependent part (the allocation fraction) which is (THIS WEEK - LAST WEEK) / ORIGINAL PRICE. That means that you could safely ignore a stock that has lost a tremendous amount of value since the start of contest that doubles in price, or worry about missing the move in a stock of adding 50% of its original price to a stock that had tripled in price since the contest start, which would only increase by the less imposing 12.5% using the newspaper “percentage”.
I have tried to in each report at least once when discussing the percentage change in a stock price to say that the number is the change in price compared to the original price, or price at the start of the contest or some such. However, it is not at all unreasonable for anyone to just pass over these words and assume that the percentages were the more “obvious” calculation. Next year I intend to have a column that gives both kinds of percentages, though selecting the big movers only using the percent change against the original price. This might cause some initial confusion, but it would be educational to see how the two numbers varied. |
