Dave Fairtex, writing for Peak Prosperity, claims he has The Smoking Gun Proving Silver & Gold Manipulation. In identifying all of the 0.5% or greater one minute price spikes over the past 6 years I’d argue he has proven how infrequent it is.
Dave’s main result is that there were a total of 135 gold events and 869 silver events where the price moved up or down more than 0.5% within a minute, with 66% of the gold events, and 54% of the silver, being downward moves. While the data shows a bias to the downside, Dave doesn’t put any of the results into context, undermining I think what is a worthwhile analytical approach.
Dave’s data set includes every 1 minute period between 4pm-3am US Eastern Time from late 2009. That is an 11 hour trading window over 6 years. The number of 1 minute periods in the data set is therefore:
(52 weeks x 5 trading days) less 10 public holidays) x 11 hours x 60 minutes x 6 years = 990,000 periods
A total of 135 gold events within 990,000 periods is 0.0136%, or 1 event per 7,333 periods. For silver it is 0.087%. That is not a lot of manipulation events and in the case of gold, Dave shows that oil, gas, wheat and corn had more events.
Dave’s other observation is that when you compare the aggregate dollar change over all manipulation events and express them as a percentage of price, gold and silver have much bigger price effects (-33% and -120%, respectively), compared to a few percent for the other contracts investigated.
However, I see two problems with this comparison. Firstly, the total dollar changes are just compared to current spot price. A $10 change on $1000 is not the same as a $10 change when the gold price was $1900. It would been more informative to have added all the percentage changes for the manipulation events so that apples were being compared to apples, in terms of price impact at the time.
Secondly, just as with the 990,000 period example, a 4% total change for Crude Oil may not be directly comparable to -33% for Gold as the normal volatility for gold and oil differ. Having a total dollar change up and down over the other 989,865 non-manipulation events would reveal the real impact these events had for the overall price change for gold and silver within the time period investigated. Given that the average gold price over the time period is well above $1150 I’d say that -33% may be overstating the case.
The other aspect of the analysis that I think can be queried is the choice of trading window, which is explained thus:
“Most goldbugs like to say that gold and silver suppression attacks occur in the “wee hours of the morning.” Loosely translated, I take this to mean during non-US and non-London trading hours. So that’s the time range I will use: 4pm-3am Eastern; from just after US market close through to the London market open.”
If it wasn’t for the fact that Comex and London do trade significant volumes, I’d say this was a US-centric “we are sleeping so it doesn’t matter” view. However, do not those same “goldbugs” proclaim that US and UK trading is all paper, compared to the “real” physical markets in Asia, using charts like this to prove their point:
Below is a table of the key gold trading hubs and their usual trading hours. The highlighted areas are when those countries are operating their trading exchanges.
To argue that 4pm-3am is “when activity is relatively lighter than usual” is to argue that the huge physical markets of China, India, Japan and Dubai don’t matter when it comes to gold and silver prices. By the way, Perth is in the same time zone as China, which means the 300-400 tonnes of gold we trade each year must also not matter.
From the table above you can see that all of the major Asian markets close down their overnight electronic platforms around 3pm New York. This is not coincidental, as that is when CME’s Globex is winding down. I would argue that a more realistic period during which the global precious metal markets are relatively illiquid would be from Globex close at 5pm (New York time) to 9pm when the SGE opens, and maybe 2:30am to 5:00am, between the China/Japan break and the setting of the AM London Price.
I have left a comment on the Peak Prosperity article asking Dave to re-run his program with a change to the trading window and accumulating percentage changes for all manipulation events as well as all 990,000 periods. Hopefully he will indulge. It would also be interesting to see the analysis run for the entire 24 hour window, to get an overall sense of potential manipulation events.