The data in all columns was ranked on a scale of 1-50.  Thus, for example, for each of those 370 sessions, a few biotechs would get a ranking of "1" for standard deviation, and a few would get "50".  The rules below can always be applied in the next trading session regardless of what the general market or biotech industry did in the previous session...you simply take current data for all biotechs, and sort it according to these rules.  If, say, we created rules like "buy any biotechs that gained more than 5% in the last trading session", it might be possible that you could scan all biotech stocks, and not find a single one that met that requirement.  With our ranked data, that's never the case.

After risk-adjustment, the single best indicator of a gain in the next session is a biotech that finished the day with a large differential between its high price and closing price.  Such an approach has averaged better than 1.3% gains per day when dividing biotechs up into 25 groups, according to the high-close differential, and outdistances the next best approach by a hefty margin.  It remains to be tested (a difficult exercise) exactly how much of this 1.3% could be captured with real money.

Combining a large high-close differential with moderately increasing volume (specifically, the 60-70% decile) over the last three days versus the last ten days pumps the daily gains up to 1.7%.

Other strong predictors of gains in the next session would be biotechs that significantly underperformed the biotech indices in the previous session, those with a low close-low differential, and those that have taken recent losses.

Indicators of losses in the next session would be biotechs that finished at or near their highs (.6% losses), or finished well above their lows.  Biotechs with weak volume in the last trading session are to be avoided.  Those with nice recent gains tend to reverse...naive momentum trading is not advised in this time span.

Combining a large close-low differential with weak volume over the last 10 days (as compared to the last 60 days) increases the losses to 1.4% per session.

Looking at non-ranked data, biotechs are to be avoided following a losing market session.  Though not particularly significant, the data does confirm that it's best to buy biotechs when the biotech industry in general has gained nicely over the last week (4.7% or more), and to a lesser extent, the last day (between 1 and 2%).

Again, a large high-close differential predicted gains.  Here, the gains were about 2.6% over five days.   Other indicators of gains were a low five day moving average and a low 20 day moving average.  Combining a low 20 day moving average with a small marketcap pushed the gains up to 4.4% per 5 day period (which didn't necessarily start on a Monday, to be clear).  

Note that the strategy of buying biotechs with large high-close differentials also figured prominently in our one-day test above.  One might see the 2.6% gains over five days versus 1.3% gains over one day as a bit of a disappointment.  After all, you get a 6.5% gain (1.3*5) if you project the 1.3% gains over a week.  However, if trading expenses are subtracted out, we wouldn't be surprised if the 2.6% gains were actually superior.

Indicators of losses included a high momentum reading (-.6%), and other indicators of strong recent gains.  Biotechs whose gains in the last trading session well exceeded the industry as a whole are to be avoided.  Our greatest losses via "dual" data were gotten by combining a low high-close differential with a volatile stock (-2%).  The data shows that if a stock has strong momentum, but was flat in the last trading session, it would probably be better to dump it (-1.9%).

Looking at non-ranked data, there was a weak tendency for biotechs to continue losing if the general market (as measured by the Russell 3000, though any relatively diverse index would probably suffice) took large losses (greater than 1.8%) in the previous trading session...buying biotechs when the market is a bit freaked is not a good idea.

After risk-adjustment, the best strategy for monthlong gains in biotechs from 1/1993 to 8/2004 has simply been to buy cheap biotechs.  Going with the cheapest 4% of biotechs has averaged gains of 6.5% monthly gains over that period.  Going with three month, one month, and yearlong losers have also been strong strategies.  Small cap biotechs also fared well.

Losing strategies included going with expensive biotechs (down nearly 1% per month), large caps, and those with large recent gains.

The period examined above was a strong one for biotechs in general.  One could argue that during such a period, it would only be natural for volatile, small-cap, and otherwise risky biotechs to fare particularly well.  So we also examined the period from 6/2001 to 8/2004, a more neutral one for biotechs, for winning and losing strategies.  The above strategies held up well, however.

Looking at non-ranked data, there does appear to be a tendency for continued monthlong gains if the previous month was a strong one for biotechs (with industry-wide gains exceeding 14%).  Weeklong biotech gains of greater than 7% also bode well for the industry.  Weeklong industry losses between 2 and 4% indicate continued losses in the industry.  Here, it seems that industry-wide momentum combined with bottom-fishing at the level of individual stocks can produce nice gains.