Skip to main content
EN
Menu

Learn the stock market › Test a strategy › Lesson 4 of 5

Costs and tax in the test

Short answerCommissions, the spread and tax are paid on every trade, so a rule that trades often loses more to them than buying once and holding. On the S&P 500 since 1928, the 200-day rule returned 6.9% a year before costs, against 6.1% for buy and hold. At 0.25% per buy and per sell it fell to 5.3%, and with 15% tax on every gain it returned 5.92% against 5.94%.

The cost per trade

The rule made 589 buys and sells in 98 years; holding made a single purchase. This is how the result changes with a cost on every buy and every sell:

Cost per buy or sellThe ruleBuy and hold
0.0%6.9%6.1%
0.1%6.2%6.1%
0.25%5.3%6.1%
0.5%3.7%6.1%

S&P 500 since 1928, without dividends, per year. Buy and hold pays the cost once.

At about 0.12% per trade the rule's lead was already gone.

The cost per trade is several things together:

  • Commission, the fee to whoever you trade through. Many US brokers charge none on stocks, as the lesson on trading costs shows, but the other costs remain.
  • The spread, the gap between the bid and the ask. It appears on no statement.
  • Slippage, the gap between the price the rule counts on and the price you actually get. A rule that trades at the close can rarely get exactly the closing price.

So a test without costs says most about rules that trade rarely.

The tax

In a taxable account a gain is taxed when you sell, and a loss can be set against gains. The rule sells often and pays tax along the way. The holder pays only on selling, and the money that would have gone to tax keeps growing in the meantime. With 15% tax on every gain, and losses set against later gains, the rule returned 5.92% a year against 5.94% for holding and selling at the end.

We used 15%, the rate the IRS says most people pay on long-term gains. A gain on something held a year or less is taxed as ordinary income, usually at a higher rate, and 93% of the rule's trades lasted a year or less. So the real tax on the rule would be higher.

What you don't see without costs

A rule that trades 6 times a year looks good in a test without costs and worse with every cost added. Rules that trade rarely lose less. When you compare two rules, give both the same cost, and test with a higher cost than the one you expect. Otherwise it is easy to fool yourself, as in the lesson on five ways to do it.

Test yourself

Two rules made the same before costs. One makes 4 trades a year, the other 40. Which loses more to costs?

Sources