Wednesday 19 July 2006

Buy in gloom, sell in boom... LVA

As anyone reading this blog probably knows, there are myriad truisms about "buy low, sell high", "buy straw hats in winter" and so forth. With most stock markets now off their highs (some more than others), this might be an opportunity to buy stocks. Especially in Australia, where a 10% correction during an ongoing bull run would be a great time to top up on your share holdings - IF we're still in a bull market phase.

But, with continuing setbacks in Iraq and Afghanistan, and the latest renewal of conflict in the middle east, oil looks like it may continue to reach record highs. We may be at the start of the next great bear market - with a global recession triggered by energy costs feeding into inflation, causing hikes in interest rates by central banks? Already New Zealand appears to be facing a need to maintain high interest rates (to contain inflation running at 1.6% per quarter) while their economy is flirting with a dip into recession. Echoes of stagflation?

Without having a crystal ball handy, it seems impossible to tell a temporary dip from the start of a descent into the abyss. There a plenty of examples of 10-15% "corrections", as well as leading into a 40+% drop during a severe bear market.

Is there some way to benefit from the known facts (a 10%-15% drop, which may just be a dip or might be the start of a bear market) without needing to correctly guess the future? One idea I've been toying with is using a leveraged or Biased version of Value Averaging [VA]. With Dollar Cost Averaging [DCA] you regularly purchase the same dollar amount each period (say $2000 per month) of a stock/mutual fund/index CFD or whatever. With Value Averaging you plan on a regular increase in the Value of your total holding and buy the amount required to bring you up to your planned amount. For example:

DCA: Purchase Total Holding
Month Price QTY Amount QTY Value
JAN $1.00 2000 $2000.00 2000 $2000.00
FEB $0.90 2222 $1999.80 4222 $3799.80
MAR $1.05 1905 $2001.30 6127 $6433.35
APR $1.00 2000 $2000.00 8127 $8127.00

Change in value of stock in period 0%
Amount invested $8001.10
Change in value of holding in period 1.57% (8127/8001.10)

This is only an rough sketch - the actual outcomes will depend on exactly how the stock price varies during the period of your investments, and the % gain isn't an annualised ROI or anything fancy.

For comparison, with standard Value Averaging, you would do the following:

VA: Purchase Total Holding
Month Price QTY Amount QTY Value
JAN $1.00 2000 $2000.00 2000 $2000.00
FEB $0.90 2444 $2199.60 4444 $3999.60 (aiming for $4000)
MAR $1.05 1270 $1333.50 5714 $5999.70 (aiming for $6000)
APR $1.00 2286 $2286.00 8000 $8000.00 (aiming for $8000)

Change in value of stock in period 0%
Amount invested $7819.10
Change in value of holding in period 2.32% (8000/7819.1)

Now, using Biased Value Averaging [BVA], you would do the following:

VA: Purchase Total Holding
Month Price QTY Amount QTY Value
JAN $1.00 2000 $2000.00 2000 $2000.00
FEB $0.90 2938 $2644.20 4938 $4444.20 (aim: $4000*1/.9 = $4444.44)
MAR $1.05 504 $ 529.20 5442 $5714.10 (aim: $6000*1/1.05 = $5714.29)
APR $1.00 2558 $2558.00 8000 $8000.00 (aim: $8000)

Change in value of stock in period 0%
Amount invested $7731.40
Change in value of holding in period 3.47% (8000/7731.4)

This is a simplified calculation where no "typical" rate of increase in holding value is assumed - in practice you would build in a realistic real rate of return when calculating the Values you are aiming for - say a real return of 8% pa for the Australian All Ords Accumulation Index.

I haven't read of this version of Value Averaging anywhere else, but it's probably not an original thought - let me know if you've seen it somewhere else, and what it's properly called (I just coined the catchy "BVA" moniker). I doubt this will pan out into a Nobel prize for economics ;)

Also, if anyone knows of a study of it's effectiveness, I'd be interested. Otherwise, I plan on doing some comparisons using historic data - maybe back-calculating using monthly data for rolling 20 year periods of the Australian All Ordinaries accumulation Index.

My theory is that using this method [BVA] you would automatically start investing in "dips" and increase the scale of purchases as the market dropped below its long term average. (You'd have use an assumption of the long term average rate of increase when calculating the total value you were aiming for each period, otherwise over time you'd be buying smaller and smaller lots until you never made another stock purchase)

I'm not sure a straight reciprocal of the deviation of the index from the expected value (as used in the example above) is best for calculating the target value for each period - I'd have to fine tune it to allow for the expected size of peaks and troughs - say 10-15% deviations above and below the long term average happening for several months every two years, and the odd bull or bear market pushing it +/- 40% from the long term average for 6-12 months every decade or so. Easiest way to "tweak" the parameters would be to back-calculate using various scaling factors. (Of course this is optimising the model for historic data - which may bear no relationship to what acutally happens in the next 20 years!).

Another consideration is at what rate you wish be investing into the market - if you are accumulating, say, $2000 per month to invest, then DCA or BVA is fine. But if you are sitting on a large sum of money to invest (say you're overweight in cash, "waiting for the right moment" to invest in stocks), then you need to consider the opportunity cost of leaving a large part of your funds in cash while you average into stocks over several years.


Anonymous said...

It's me the first post bandit. This is my first time to your site, so the first post seemed appropriate =) just wanted to say hi and great blog!


lostcowboy said...

Hi, best description of value averaging I have found on the web is at,

At the bottom of the page is a link to the spreadsheets from the book.
You may also want to do a search on the word synchrovest. Synchrovest is a very interesting plan from R. Lichello.