Let's kick off this EOD Trading category with an example (and warning of sorts) of what can happen when using dividend adjusted data when following signals live - whether it be live trading, paper trading, or WealthSignals trading.
For the sake of a simple example, let's say your strategy buys 3% below today's close, i.e., Close[bar] * 0.97. And, your Data Provider is Yahoo! Finance with dividend adjustments enabled (see Yahoo! view in Data Manager).
Your strategy Alerts on 3/12/2013 to buy GRMN at 34.629 the next day. As it turns out, the low of GRMN was 34.62 on 3/13/2013, so you would have a position with WealthSignals.
However, GRMN paid a $0.45 dividend on 3/13/2013, the ex-date. Consequently, when you update your Yahoo! data after the 3/13 close, you'll still see that GRMN's low for the data was 34.62, but the strategy will not be holding an open Position. Why?
Because, applying that dividend to the data changes (lowers) the prices of all bars prior to the ex-date by multiplying by the dividend adjustment factor: 1 - dividend/close. (Note that Volume is divided by this factor.) The strategy uses the new data when calculating the limit price for 3/13, and instead of 34.629, it changes to 34.1925, which is below the low of the bar, so no trade.
As a consequence you have an "orphaned Position", i.e, a Position that you're holding in a live or WealthSignals account, but one that your Strategy is not aware of.
First and foremost, I recommend using the Trades+ Visualizer. It has an Open Positions view that you can use to cross check your live/WealthSignals Open Positions with those of your Strategy. If they don't match, you have to investigate in order to get back into synch.
I won't recommend that you do not use dividend-adjusted data for live trading because a dividend payout when not back adjusting has an artificial negative impact on price. In other words, GRMN dropped over 1% on 3/13 simply because 45 cents per share was shaved off and distributed to shareholders. Adjusting for dividends accounts for that, and is more representative of your backtest strategy.
Nonetheless, you're less likely to get out of sync by disabling the dividend adjustment for live trading; with the side-effect of trading Positions that your system wasn't necessarily designed to trade.
Another way to approach using dividend adjusted data would be to account for "tomorrow's" ex-date payouts in your strategy signals/alerts. With access to the dividend stream, you could do this fairly easily with backtesting by "legally peeking ahead". (Dividends are announced well in advance of the ex-date, so that's why it's "legal" peeking.) However, dividends are not available in Yahoo! until data are updated for the ex-date, so the problem still can't be avoided with live trading without a manual effort.
As a note, U.S. customers using Fidelity data have access to future dividends, but Fidelity does not provide the back-adjustment option.
Re: Trades+. What if we let the WealthSignals Publisher handle it? We would need some logic to match the open positions between a WealthLab backtest and a corresponding WealthSignals trading system, and act accordingly e.g. highlight the mismatched Signal not backed by a WealthSignals position.
As an an aid for WealthSignals, I think that would be a fine enhancement.
However just to be clear to the casual observer, the trading system/account synch issue exists for any [primarily EOD] trader who uses dividend-adjusted data.
The same may apply for stock splits on the day of the transaction. Day before the split, a signal is generated at the pre-split price. Day of, that price will more than likely execute at the pre-split signal's price.
Regarding dividends: Accounting-wise, Wealth-Lab script's Trade history should list the dividend amounts and dates, so we can confirm dividends are being applied to our Wealth-Signals systems. This also means listing dividends in the trade history on the Wealth-Signals site. I could not find an example on the Wealth-Signals site.
Total accumulated dividends and Dividend Annual Rate of Return statistics would be nice, to allow dividend investors a means to evaluate a system utilizing dividends.
Thank you for the suggestions. Dividends are undoubtedly being applied to WealthSignals systems, but evaluating their impact on a system's performance seems important.