Desperado
Member
A Perfect Display of Why Market Timing Doesn’t Work
A group of 48 Federal Employee Thrift Savings Plan (TSP) investors tracked their returns from Jan 1 2006 to Nov 24 2006 at TSPtalk.com. These investors changed their asset allocations on average 37.6 times over the 11 month study period, with the most frequent trader swapping 157 times, (impressive given the fact that the NYSE was only open 226 days.) As the TSP is a tax-deferred, no-fee plan these traders incurred NO transaction costs. Their investment options include 4 very low cost index funds which mimic the S&P 500, the Wilshire 4500, the EAFE, and the Lehman Bros Bond Index, and a special fund called the G fund, which essentially has the return of a medium-term government bond fund with the risk of a money market fund. There are also 5 “lifecycle” funds within the TSP program, but these are infrequently used by these investors. Transactions requested prior to noon Eastern time take place at the close of business that same day.
The individual fund returns for the study period were:
G Fund (Treasuries): 4.48%
F Fund (Lehman Bond): 4.50%
C Fund (S&P 500): 14.17%
S Fund (Wilshire 4500): 15.43%
I Fund (EAFE): 21.15%
The mean return for these investors was 11.64%. The median return was 11.97%. The best return was 22.16%. The worst return was -4.01. The standard deviation of returns was 5.45%.
Comparison against appropriate indexes would be appropriate:
Perhaps the best comparison for these relatively aggressive investors would be against an all-stock “Total Global Market” portfolio (50% I, 39% C, 11% S), which bested 92% of them. Even an all-stock buy-and-hold investor who feared foreign markets and invested only in the “Total US Stock Market” portfolio (78% C, 22% S) would have beaten 35% of investors. Only 52% (not statistically significantly different from ½) of investors topped a “know-nothing” portfolio of 20% in each of the 5 funds. Even the average lifecycle fund investor (who often held a large percentage of relatively low-performing fixed income) bested 46% of these traders.
But what about the “Warren Buffets” of the group. Surely there must be at least a few very talented traders in this group. In a random distribution, one would expect approximately 2.5% (or about 1 trader) to perform better than 2 standard deviations above the mean. In fact, no traders managed this feat, suggesting that the outperformance of their top performers was less likely to be due to skill than sheer luck. In fact, a distribution of monkeys throwing darts to choose their allocations would likely have produced a higher performer than these 48 investors. One would also expect 16.9% of investors to perform better than 1 standard deviation above the mean. In fact, only 14.5% of investors did so.
It is noteworthy that these investors were functioning in a “zero-sum” game, quite different from the typical investor, in that they pay no portfolio fees (aside from the miniscule ERs of the individual funds), no transaction costs, and no taxes on gains. If these traders had been operating in a typical investing environment, their transaction costs and taxes would have made their performance even more dismal.
The correlation coefficient between # of trades and return was 0.048, which means there was no significant benefit to trading more often.
A group of 48 Federal Employee Thrift Savings Plan (TSP) investors tracked their returns from Jan 1 2006 to Nov 24 2006 at TSPtalk.com. These investors changed their asset allocations on average 37.6 times over the 11 month study period, with the most frequent trader swapping 157 times, (impressive given the fact that the NYSE was only open 226 days.) As the TSP is a tax-deferred, no-fee plan these traders incurred NO transaction costs. Their investment options include 4 very low cost index funds which mimic the S&P 500, the Wilshire 4500, the EAFE, and the Lehman Bros Bond Index, and a special fund called the G fund, which essentially has the return of a medium-term government bond fund with the risk of a money market fund. There are also 5 “lifecycle” funds within the TSP program, but these are infrequently used by these investors. Transactions requested prior to noon Eastern time take place at the close of business that same day.
The individual fund returns for the study period were:
G Fund (Treasuries): 4.48%
F Fund (Lehman Bond): 4.50%
C Fund (S&P 500): 14.17%
S Fund (Wilshire 4500): 15.43%
I Fund (EAFE): 21.15%
The mean return for these investors was 11.64%. The median return was 11.97%. The best return was 22.16%. The worst return was -4.01. The standard deviation of returns was 5.45%.
Comparison against appropriate indexes would be appropriate:
Perhaps the best comparison for these relatively aggressive investors would be against an all-stock “Total Global Market” portfolio (50% I, 39% C, 11% S), which bested 92% of them. Even an all-stock buy-and-hold investor who feared foreign markets and invested only in the “Total US Stock Market” portfolio (78% C, 22% S) would have beaten 35% of investors. Only 52% (not statistically significantly different from ½) of investors topped a “know-nothing” portfolio of 20% in each of the 5 funds. Even the average lifecycle fund investor (who often held a large percentage of relatively low-performing fixed income) bested 46% of these traders.
But what about the “Warren Buffets” of the group. Surely there must be at least a few very talented traders in this group. In a random distribution, one would expect approximately 2.5% (or about 1 trader) to perform better than 2 standard deviations above the mean. In fact, no traders managed this feat, suggesting that the outperformance of their top performers was less likely to be due to skill than sheer luck. In fact, a distribution of monkeys throwing darts to choose their allocations would likely have produced a higher performer than these 48 investors. One would also expect 16.9% of investors to perform better than 1 standard deviation above the mean. In fact, only 14.5% of investors did so.
It is noteworthy that these investors were functioning in a “zero-sum” game, quite different from the typical investor, in that they pay no portfolio fees (aside from the miniscule ERs of the individual funds), no transaction costs, and no taxes on gains. If these traders had been operating in a typical investing environment, their transaction costs and taxes would have made their performance even more dismal.
The correlation coefficient between # of trades and return was 0.048, which means there was no significant benefit to trading more often.
Last edited:
