A.I.M. Users Bulletin Board

[ls7550]

Hi Allen

It probably doesn't really matter. I wonder if Lichello did it the same way as Newport.

Consider a log stochastic measure

( current - bottom ) / ( top - bottom ), log scaled

For example with the Dow at around 17,500 I might mentally consider that if prices halved and halved again I'd like to be all-in (0% cash), or if the price doubled and doubled again I'd like to have sold all-out (100% cash).

Bottom = 4,375 (halve and halve again down from current 17,500 level)
Top = 70,000 (double and double again up from current 17,500 level)

Current indicated cash reserve (log stochastic)

= ( ( log(17500) - log(4375 ) ) / ( log(70000) - log(4375) )
= 0.5 (50% cash reserve) indicated.

Adjust cash reserve as indicated as the share price moves around (i.e. leave top and bottom figures as-is and revise the current value) and that produces a different log stochastic value (indicated cash % reserve) - you're Martingale betting stocks (adding after a loss, reducing after a win).

Anther interesting mathematical 'game' is d'Alemberts betting sequence. Start with one unit stake and after each losing play increase the stake by one unit, after each winning play reduce the stake by one unit (minimum one unit stake).

Stake $1 on a coin flip, lose,
increase stake to $2, lose
increase stake to $3, win - and you're at break-even despite 2 losing plays, one winning play.
Reduce stake to $2, win, and you're up overall ($8 total laid in stakes, $10 returned from winners, $2 overall profit), despite there having been equal numbers of winning and losing plays (50/50 outcome).

In a Casino martingale (d'Alembert) ultimately fail due to eventually encountering a long sequence of losing plays that you're unable to fund. For stocks depending upon how you partition up a single 'play' there are potentially a finite (maximum) number of losing plays and you can bankroll accordingly so that even under that worst case you're still able to fund positions.

The decision you have to make (that can be avoided by using a guide such as AIM) are the top and bottom levels. If you're guess proves to be 100% accurate you'll maximise rewards. If your guesses were wildly out you detract from rewards. AIM's guess for the top is a infinitely high value, it never sells out of stock completely. Its guess for the bottom is broadly around a price decline comparable to the percentage cash value (50% cash covers a 50% decline in share prices before being exhausted). As such AIM isn't the most efficient overall, its top figure guess is too high, and in some cases its choice of bottom figure is also too high. But as a somewhat simplistic guide its generally OK.

So don't bother worry about the exact calculations and following AIM to the last penny, its more a case of the overall 'artful' concept that matters rather than the exact/precise scientific/mathematical measures mattering.

Don't let AIM become a Microsoft Office, rather let it be more like LyX. With MS Office you end up concentrating more on the formatting/appearance than the words; With LyX you're steered towards concentrating more on the words than the formatting. Your job as a AIM manager is to decide whether you'll just follow AIM's advice and trade at set intervals, or whether you'll apply some logic to what AIM is indicating and perhaps delay trading, or trade earlier (between) set review points. If those guesses prove to be accurate you'll enhance rewards, guess wrong and you'll relatively lose out.

Deciding upon AIM settings and what to feed AIM is the most difficult and 'speculative' element. Often results may prove to have been no better/worse than had you just AIM'd the market (index) using set/periodic reviews. AIM doesn't have a answer to that to consistently improve rewards. That's down to you to decide, picking whatever stocks/holdings you perceive to have a better prospect of being the more rewarding compared to other choices.

Myself, I'm migrating major holdings increasingly towards holding just a few investments, international stocks, domestic stocks and bonds that I anticipate each individually having above average prospects whilst also having below average downside risk and I align exposure periodically (relatively infrequently) to a vWave type measure as/when it looks appropriate to do so (potential turning points or weightings have become excessively out of kilter).