I wanted to update some of my findings last night, but due to software difficulties, the analysis took hours instead of the 15-20 minutes I anticipated. My apologies for the delay. Dimitri and I pored over the results of our preliminary quick tests, searching for parameters that might enable us to distinguish the winners from the losers. We found several parameters that seemed to separate the men from the boys. Unfortunately, these parameters are specific to the research program that I use and if I mention them, they won’t have any meaning to you unless you’re using the same software. What I can say is that in general, the cruddiest, lowest-priced stocks turned out to be the biggest winners while the more attractive, higher-priced stocks were the biggest losers. Who knew?
After modifying our search criteria to include our new parameters, we back-tested it through all Up market periods beginning from 9/7/04 to 11/1/07. (We went into cash during the down cycles.) Our search criteria were the following:
1. Maintain a portfolio of at most 10 stocks.
2. When a buy signal is triggered, buy stocks at the next day’s average price. (I don’t like buying at the open.)
3. No duplicate buying of current holdings.
4. No stop/loss was set which means that stocks were held for the entire Up cycle, even if the price dropped to zero.
5. 50% margin account used.
6. $9.95/trade commission fee.
7. Profits/losses were added into the buying power of the next Up market cycle (i.e., the buying amounts were not fixed).
The results were staggering. Beginning with just a $5000 investment in a margin account, the final total at the close on 11/1/07 would have been $2,252,882 (including account interest) for a return of 44,958%! But of course, when things are too good too be true, they usually are. Looking at the actual trades, we found that the number of shares that were bought by the computer greatly exceeded the 65 day average volume and probably exceeded the entire stock’s float.
Out of a total of 33 stocks purchased during the entire time, there were 9 winners and 24 losers. However, the size of the winners squashed the losers. The biggest percentage gainers were the following:
MRFD: 154,250% on 272,000 shares (avg. vol. = 3100 shares) (price went from 0.001 on 9/7/04 to $1.60 on 6/1/05. This stock later peaked at $13 on 2/20/07–WOW!)
AWTI: 1949% on 4,048,000 shares (avg. vol. = 20,000 shares)
NSTLQ: 698% on 4,129,000 shares (avg. vol. = 18,500 shares)
MXDY: 649% on 9,090,000 shares (avg. vol. = 34,000 shares)
USXP: 600% on 105,181,510 shares (avg. vol. = 95,000,000 shares! This trade may look feasible, but its chart suggests that it’s the market makers who are creating this volume and it’s the market makers who are probably the only ones making money.)
The largest losers happened to be the only stocks whose beginning prices traded over $1:
SVXA: -94% (from $150 to $10.50)
ADSD: -89% (from $26.25 to $3.15)
Fortunately, both stocks were bought early on with a small amount of capital so their impact on the overall portfolio was minimal. However, if I were to actually trade these, I’d certainly set stop-losses (10% for SVXA and 20% for ADSD). Since the majority of the stocks trade in millicents, setting stop-losses on them is rather ridiculous.
So, what did we learn from this exercise? Plenty. Penny stocks can be very lucrative but they are fraught with inherent drawbacks, the obvious one being the limitation on the number of shares available for purchase. One also has to remember that trading stocks on the pink sheets involves going through market makers where for one, the bid/ask spread maybe large (as is probably the case with USXP), and two, you might need special permission from your brokerage firm to trade these. However, if you have some extra dough that you don’t mind losing, then you might get lucky and strike it big–big enough to fund your retirement account or send your kid to an Ivy.
I do think this approach has some value that requires further examination. What Dimitri has suggested is that we go back through our previous simulation and purchase the same stocks with the number of shares totaling at most half of the average daily volume. As long as our simulation recommends the stock, we’ll be adding to its position (instead of buying it all in one lump) on a daily basis. That’s a much saner way to do it, but it’s going to involve a lot of input by hand and many hours of toiling over a hot computer.
I hope Dimitri isn’t planning on watching the Super Bowl…
Addendum: I know you’re probably chomping at the bit to see what stocks my model is pulling up for today. Here’s the top 20 of the 37 stocks that came up:
SNVH, TGLI, OGOH, PGSW, CIGI, GAMT, UDTT, SGRZ, TRNP, PWTC, WLKF, CKEI, PMED, ILVG, MMAM, WHAIQ, SSWME, SMTR, SLVO, TRPH
Posted by Dr. Kris at 1:08 PST