In response to the original topic. I apologise if this lands up in the wrong discussion/topic as this is my first post here (please feel free to move it to the appropriate thread).
)-------
Out of personal experience, I’d like to share some technical market timing models that I have come across. All timing models are based on 2 main drivers - momentum and deviation from the mean.
You basically are attempting to do one of 2 things from any timing model a) Cut down asset class exposure/rebalance among asset classes or b) Hedge exposure by using derivatives. The trade-off is usually capital gain at risk combined with transaction costs (cap gain taxes, market liquidity of individual positions etc).
The models used by most sell-side houses and institutional investors tend to be broad based in terms of input data. For example, Morgan Stanley Europe uses a CVI model (composite value indicator) that takes into account equity markets P/E, bond yields, macro fundamentals (ISM, payrolls, unemployment figures), investor risk appetite (option put/call ratios, open interest) and so on; signals are generated based on deviation from the mean. You can go through this MS presentation to understand the model better.
For individual investors, given these data are costly to obtain, a simple momentum based model works quite well. The starting point is to decide whether you want to preserve capital by selling off or hedging.
The most simple momentum model is a moving average model which shows you the trend and slope of the trend (very important) in relation to the current spot price of the asset. Almost all investors pay attention to the 200 day simple moving average (SMA) so this would be a good place to start; this works best for the index as opposed to individual stocks so you eliminate event risk.
Moving average models are lagging indicators so they will not flag off a market top/bottom, they simply show you the trend. Longer moving averages aim to eliminate noise from the trend and gives less weightage to recent price action.
There have been studies done in the US (http://papers.ssrn.com/sol3/papers.cfm?abstract_id=962461) that backtest this in the US since 1900. A 200SMA model sacrifices the upside in bull markets (by being late to the party) for entirely avoiding long protracted bear markets (notably 1929-32 and flagged the 2001 bubble just after it’s peak).
A frequent problem is that these models generate false positive signals, more often than not, called whipsaws (usually a dead cat bounce in bear markets or steep sell off’s in bull markets). You can refine these models to make them more robust. But I need not remind forum members that derivative exposure combined with whipsaws can be lethal if you are not disciplined.
I believe these models are beneficial for fundamental long-term investors by pointing out flags before it’s too late. People usually compare it to buy-and-hold strategies but for stock pickers who aim to generate alpha, this is an invaluable tool.
I am working on some timing models for the Nifty and hedging using Nifty futures. It is currently technically driven (ie moving averages) and the backtest is promising (90% hedging achieved with significant MTM volatility). I’m working on making it more robust to minimise whipsaws as well as incorporating fundamental valuation based flags.
I would be happy to share them with members on this forum and work collaboratively to refine the model if there is interest.
Cheers.