A Fibonacci-based model of the gold price



A Fibonacci-based model of the long-term gold price

by Mestre Sócrates

What is the deal with gold? Is it a ‘barbarous relic’ as Keynes claimed (mendaciously, as it turns out, since when asked on the eve of Bretton Woods to design an international currency, he proposed BANCOR: an international currency backed by... gold!), or are we assisting a titanic struggle between corrupt central banks and the forces of sound money which will eventually triumph, sending gold to a price of 8 squillion dollars and condemning everyone to economic hell except for the faithful few who ‘got gold’ in a kind of financial version of ‘the rapture’?

I suspect that the truth is rather more prosaic: Gold is just a passive proxy for ‘stuff’ as opposed to fiat money and hence it tracks the real inflation rate for ‘stuff’ since its price represents a kind of market’s best guess as to the expansion of the broad money supply. Everyone complains that their true cost of living increases by much more than a manipulated government inflation index which only tracks inflation for things whose price goes down (like mass-produced high tech) or which don’t actually exist (like rent which home owners pay themselves) . Hence, it should be the most obvious thing in the world for the gold price to rise (and carry on rising by 15-20% per year and the only real conspiracy pushing the gold price is the universal ‘Hyperkeynesianism’ of governments which believe that they serve the greater good by creating money out of thin air in excess of the underlying real rate of economic growth.

We thus find that the CAGR for the gold price since 2001 has been around 19% (if we take a 3-period moving average of the average annual prices of gold to smooth out the swings, we find gold went up by 10-14% per annum in 2003-2006 and has been rising by 17-21% per year over the last 4 years. If you had to bet on the likely rate of gold appreciation over the next 4-5 years, the evidence would thus lead you to say 25% rather than 50% or 100%.

The above may sound awfully boring to gold traders hoping to retire on the coming “moonshot” in precious metals prices but a) having an accurate conceptual model of what drives the gold price will protect your portfolio from the vicious downtrends that the ‘true believers’ have to endure and b) even if the inflation rate accelerates by 3% per year (i.e. from 19% to 22% to 25%, etc.), taking an approximate current price of $1100, we are still on track for $2600 gold by 2014.

Indeed, gold appears to have a predictable trading pattern of a new high, a slam down to the previous fib level, reworking back to the previous high, a dull 6-month ‘handle formation’ period and then a 2-3 month period of rallying to a new fib level. This has given workable projections for the gold price years in advance.

I commented on this long-term cup and handle pattern on the gold chart to Rick and he asked me to write up my comments for public consumption.

These fib levels are given by successive half-roots of the golden number (φ = (1+√5)/2) (i.e. φ-3, φ-2.5, φ-2, φ-1.5, φ-1, φ-0.5, φ0, φ0.5, φ1, φ1.5, φ2, etc.) which may be approximated by a familiar series of real numbers: 0.236, 0.382, 0.486, 0.618, 0.786, 1, 1.272, 1.618, 2.058, 2.618, 3.330, etc.) If we calculate these ratios for the long-term downswing in the gold price from 850 to 253, we get a sequence of prices (…, 321, 340, 364, 394, 432, 481, 543, 622, 722, 850, 1012, 1218, 1482, 1815, 2241, 2782, etc.) (i.e. 722 = 253+(0.786*(850-253))

I studied these Fib numbers about 10 years ago with Larry Pesavento, before they became fashionable. One of Larry’s ‘big ideas’ was the particular significance of the 0.786 level, which marked the transition from a simple retracement to a primary bull trend – furthermore, once breached, a price could take out the 1.00 level and go straight to 1.272. The last major hurdle was to break through the 1.618 level and then ‘the sky was the limit’. This applied to any financial instrument.

How does this apply to gold? (Unfortunately, I have to use London fix numbers as it’s the best price series I can find, but they work quite well.)

On the gold retracement, the first major ratio we look at is the 0.238 ratio at $393. On 5/2/03 – the LF (London fix) was $382.10 – Gold began a 2-month decline to $319.90 on 7/4/03, recovered to $371.40 on 27/5/03 (2-months), went down to $342.50 on 17/7/03 (close to a 61.8% retracement) before printing $382.25 on 9/9/03 (it then meandered in a $20 range before breaking up definitively in November.

We then had 2 years of rather messy patterns, but note that there were oscillations around the .382 level ($483) in Sep-Oct 05, then another oscillation around the .486 level ($545) in Jan-Feb 2006 before gold shot up to the .786 level ($725) on 12/5/06. We then started a major decline with 2 bottoms, $569.50 on 15/6/06 and $560.75 on 6/10/06.

At this point, we started to rally, reaching $685.75 on 26/2/07 and $691.40 on 20/4/07, followed by a long and boring consolidation which went as low as $642.10 on 27/6/07, before finally taking out the old $725 high on 19/9/07. The C&H pattern played out over 16 months.

We then rallied for 2 months to the 100% target at $850 and spent 2 months oscillating between $790 and $850, before moving higher to peak at $1011.25 on 17/3/08 (close to the 1.272 target at $1014). This is a classic 0.786 breakout to the 1.272 level.

The rest is history, with a collapse in the market back to twin lows of $712.50 on 24/10/08 and $713.50 on 13/11/08 (London fix).

This was followed by a rally which again oscillated around $850 in Dec 08-Jan 09, before reaching $989.00 on 20/2/09, followed by a 2-part handle formation which finally broke above the old high on 4/8/09. The C&H pattern played out over 18 months.

The last rally stopped at $1212.50, on 2/12/09 – close to the 1.618 target at $1218.

We can see from the above that when gold approaches an important fib level, it either oscillates for a couple of months and goes through to the next fib level or starts a decline which turns into a C&H formation which to date has taken around a year and a half to play out.

The recent move down from $1212 to $1058 (LF) is far too big for a $50-60 oscillation, so we can surmise that gold is doing another C&H routine. Where is the low? Since $1014 was a real resistance level, it appears to be more likely to be closer to the final low than the next fib number down at $850.

Interestingly, if we look at the microstructure of the initial sell-off from the Mar 08 high, we see that gold initially bounced off $853 convincingly on 1/5/08 and then rallied back to $986 on 15/07/08. What came next was the massive sell-off down to below $681 (spot), but if we had followed the above model, we would have sold anyway at $986, since we would have expected an utterly dull 6 months of handle formation.

The behaviour around $725 and $1014 suggests that when gold peaks and starts to form a C&H, it then takes 5-7 months to reach the bottom of the cup.

Will gold repeat this relatively precise pattern in the future so in the future? Who knows? But if gold were to take out $1215 tomorrow – this would still tell us something, namely that fiat money growth was out of control and that inflation was accelerating. This is such an important point that it is worth repeating – the point in time when gold bottoms relative to the model’s prediction (April/May) is giving you an insight into where we stand in the deflation/inflation debate.

On the other hand, all other things being equal, the model tells you that during this initial 5-month bottoming phase you can trade gold but should be wary of investing in it unless you’re relaxed about the possibility of a 10-15% drawdown. Gold is just ‘doing its thing’ but from April/May onwards should again be generously inclined towards longer-term investors and if it prints $1014, then it is a signal for a great $200 trade, almost as good as the breakout above $1225 when it finally happens.

It also gives you a roadmap for the next 18 months: If history is to repeat itself, gold has to print $1014 on the London fix as a bottom signal, but this is due no later than May 2010 (it may dip $20-30 below this intraday, but it’s the London fix which counts). May-Sep will give a nice rally to around $1190, followed by a retrace to around $1110 in December, with $1210 taken out in March 2011 and the fast run to $1490 around May 2011. I am merely projecting the previous time frame here and do not claim to have a crystal ball, but would like to say that I would have done much better in my trading than was the case had I listened to this model 6 months ago.

Mestre Socrates: written on 2/17 and revised on 3/30/2010.

................
................

In order to avoid copyright disputes, this page is only a partial summary.

Google Online Preview   Download