Backtesting and adjusted close

Discussion in 'Strategy Development' started by trade4ever2day, Sep 29, 2010.

  1. I understand that when backtesting stocks one needs to adjust the series for any splits but I do not understand the adjusted close in the case of dividents. Do you people adjust the price series for dividents? I mean I understand the issue of a gap created by stock splits and why a price series must be corrected for that before backtesting but I do not see why there should be adjustments for dividents. Any thoughts?
  2. This should be taken care of by your data provider.
    Unless of course you use free data such as yahoo, etc
  3. m.runner


  4. Interesting. Thanks. I think the OP is asking whether adjustments to the data for dividents are required for proper backtesting. I gather that when a stock splits there is a proportional increase in historical volume because 2 new shares correspond to 1 old share, in the case of a 2:1 split for example. As a result, an adjustment to the data is required to take care both of the gap but also of the volume. What about dividends though?

    Let me put it in another way. Data are adjusted for splits to maintain continuity in the series. If a stock trades at $40 and splits 2:1 then instantly the price becomes $20 but oustanding shares double. If you have an open trade at $40 for 100 shares your broker will convert that to 200 shares for $20. But if you have a trade at $40 for 100 shares and the stock pays $1 divident, you still have $100 shares at $40 open price plus $100 proceeds. I guess the question of the OP is then, if I understand it correctly, do you adjust the previous data for the divident? Historically speaking, your entry price will still remain at $40, it will not change, where in the case of a split, the price in the books will change and that is why you have to adjust. So I conclude, it may not be entirely correct to adjust historical stock data for dividents.
  5. jharmon


    Index providers do not adjust for normal cash dividends. They do adjust for special dividends though.

    The same goes for Bloomberg and I believe Reuters too.

    A better method for backtesting in general would be to:
    a) adjust as per Bloomberg and Reuters; and
    b) also tally up dividends as they apply to your trade

    On perspective: You can consider that dividends are just a normal cost to the business - just like the CEO's salary. You don't adjust a company on the basis of its costs...

    Most normal cash dividends on US stocks are small enough to be neglible anyway.