Three years ago on November 18th a landmark event polarized the gold world. The first gold exchange-traded fund to trade in the United States, the StreetTracks Gold Shares, was launched. Although born in controversy, even GLD's original detractors cannot argue with this ETF's stunning success since.

Nearing its third birthday, GLD just reported massive holdings of 19.3m ounces of gold held in trust for its investors. When your gold holdings get this large though, you don't even measure them in ounces anymore. Instead you switch to metric tonnes. This week, GLD reported nearly 600 tonnes of physical gold bullion under management for American investors. This is a staggering amount of gold.

If GLD was a central bank, it would nearly make the top 10 in the world for gold holdings. This single American gold ETF has more gold than 100 individual central banks. GLD is just a hair away from overtaking China too, which would give it the 10th spot. This ETF has more gold than the individual (not collective) central banks of Spain, Russia, India, Venezuela, the UK, Saudi Arabia, and South Africa!

According to the IMF, total global official gold holdings in late 2006 ran at 30,435 tonnes. It took central banks decades, and in many cases centuries, to accumulate so much gold. Yet after three short years, the blink of an eye compared to millennia of gold accumulation, GLD's holdings are nearly 2% of those of all the world's central banks combined! And if central banks are radically over-reporting their holdings as some suspect, GLD may very well hold a substantially higher fraction of above-ground gold.

While I will probably never buy GLD myself as I happen to be a physical gold guy, I am very excited about its dazzling success. Over two years before GLD launched, I was writing about how awesome a true gold ETF would be for the young gold bull. It is not that investors need a replacement for physical gold coins, we don't. GLD wasn't created to attempt to replace gold, but instead merely to track it.

I strongly believe every investor should have a foundation of physical gold coins in his own immediate physical possession before he ever buys a single paper investment. Real physical gold is the ultimate insurance policy to protect your hard-earned wealth from devastating low-probability events that periodically ravage paper markets. It is the one timeless asset that will thrive through all financial storms.

Unfortunately though, only a very small fraction of American investors own physical gold today. Most have simply not yet "discovered" gold and realized how strong its bull has become. Others know of gold, but either cannot buy it physically for contractual reasons or are just not motivated to find a local coin store and buy some. Still others understandably balk at the gigantic (by stock-market standards) markups that coin stores need to charge to stay in business.

GLD is not for us hardcore physical gold investors, it is for these people. GLD allows an investor to effortlessly add a little gold-price exposure to his portfolio. GLD enables institutional investors that can only trade in the stock markets per their charters to get gold exposure. GLD acts as a gateway for new investors to add "gold" to their portfolios without spending the time to find a good coin store. GLD gives new investors time to get comfortable with gold investing before they finally decide to buy physical gold.

GLD is perfect for speculators too. They can buy and sell it at will instantly with their stock accounts, no travel or hauling physical gold around is required. Sooner or later, I believe a robust options market will spring up around GLD too, offering even more gold-exposure trading alternatives for speculators. And since the GLD ETF is traded just like a stock, its bid-ask spread is a tiny fraction of those found in coin stores. It is a very efficient way to trade gold.

GLD creates a critical conduit that enables the vast pools of capital in the stock markets to flow into physical gold bullion. If stock-market demand for GLD exceeds the underlying demand in the physical-gold markets, the GLD trust goes out and buys gold bullion to shunt this excess stock demand into gold. If GLD did not equalize stock demand into physical this way, soon its price would decouple from gold's to the upside. This would obviously be unacceptable for an ETF explicitly created to track the gold price.

But of course feeding stock-market capital into the gold market is a double-edged sword. If selling pressure on GLD is greater than that in gold, the ETF's custodians have to sell gold bullion to vent this imbalance into the physical gold world. If they did not do this, GLD would decouple from gold to the downside. So GLD and the other gold ETFs will increasingly contribute to gold volatility as their holdings grow.

The only way any ETF, including this one, can consistently and precisely track its underlying asset is if the ETF's custodians rapidly equalize supply/demand differentials that arise between the ETF and the asset it tracks. GLD and the other gold ETFs are so awesomely bullish because they create the mechanism, the conduit, for the vast pools of stock-market capital to flow in and out of the physical gold market.

And despite GLD's massive gold holdings, it is still vanishingly small compared to stock-market capital. With $16b worth of gold under management, GLD is still only 1/13th the market capitalization of Google alone! Compared to the entire S&P 500's huge $14,003b market cap, GLD's trivial value looks like a rounding error. So should we gold investors want a way for stock capital to flow into our tiny market? You're darned right we should! The more capital coming in, the bigger this gold bull.

GLD's impact on the gold world has already been big in its initial three years, and it will only grow as GLD becomes more widely known as the premier way for American stock capital to game the gold price. The most intriguing characteristic of GLD in its young life is its gold holdings relative to the gold price. As this chart shows, so far GLD has contributed far more to gold's upside moves than downside moves.

After its birth in November 2004, GLD's holdings rocketed from 8 tonnes to 100t in its first week of trading. Obviously there was plenty of pent-up demand for gold-price exposure for stock-market capital. But intriguingly, GLD holdings even continued to grow during the gold consolidation of the first half of 2005. This means demand growth for GLD, despite the weak gold prices, exceeded that of gold itself. The ETF custodians had to equalize this excess stock demand into gold by buying more bullion.

By late 2005 and early 2006, gold was soaring in a powerful upleg, its first since GLD was launched. GLD bullion holdings rocketed in sympathy, blasting 105t higher to 334t in December and January alone! Once again GLD holdings were growing at a much higher rate than the underlying gold-price appreciation. Since demand growth for GLD far outstripped that of gold this excess demand was shunted into the metal.

Then in February 2006 gold started consolidating, but interestingly GLD's holdings remained stable. When GLD can track gold without buying or selling bullion, it means that the individual supply-demand balances of both GLD and physical gold are essentially equal. So despite gold's early 2006 consolidation, the selling rate of GLD did not exceed that of physical gold.

A couple months later in spring 2006 gold rocketed higher in a terminal parabola, its sharpest run of its entire bull market. GLD holdings were pretty stable during this parabola, indicating that its own supply/demand profile matched that of gold's closely. As long as the GLD ETF closely tracks gold's underlying price moves, its custodians do not have to buy and sell gold bullion.

Right after this gold parabola topped, gold crashed in mid-2006. Selling in the physical and futures markets was fast and furious and the gold price plummeted. Naturally GLD fell too as its own holders fled the gold carnage. But interestingly as you can see above, the GLD bullion holdings only had a trivial dip through this crash. The stock guys who owned GLD were no more scared than the gold traders so they didn't dump their GLD at a much faster rate than gold itself was falling.

And as gold drifted lower following a bounce after that mid-2006 crash, GLD holdings were stable and even grew modestly. As gold started rallying again into early 2007, GLD again caught the imaginations of stock investors and they bought GLD at a faster rate than gold was being bought. So the custodians once again shunted this stock demand into physical gold by adding more bullion. While gold's weakness in Q2 this year weighed on ETF holdings a bit, they soon shot to new highs as our latest powerful upleg began.

And now, three years after its launch, GLD has soared to the staggering 600t mark. Since many existing gold-coin investors still view this "paper gold" with disdain and suspicion, I don't think it was traditional gold capital that bid GLD up to 600t of gold. No, more than likely it was new stock capital that hadn't yet been active in this gold bull in a meaningful way.

Pick a number, but I'd bet that at least 80% of GLD's run higher was driven by non-traditional gold investors. This conservative estimate works out to 480t. So thanks to the mere existence of this flagship gold ETF, somewhere between 480t to 600t of gold made it into the portfolios of American stock investors that probably wouldn't have otherwise. And GLD is but one of many gold ETFs worldwide!

And the fact that GLD's holdings have spent three years growing on balance regardless if gold is soaring, sinking, or drifting is extremely bullish. It suggests that there is a vast untapped market of capital that trades in stock accounts that wants some exposure to this gold bull.

The more that mainstream investors new to gold become aware of it, the higher demand for GLD will grow. Awareness should ramp rapidly as gold hits new all-time nominal highs that the media will cover extensively. And as long as GLD demand growth outstrips gold's, this ETF will drive gold higher and faster than the metal would otherwise achieve without stock capital chasing it. Shunting stock investors' capital into physical gold via ETFs is a very, very good thing for all gold investors.

GLD's daily trading volume is also intriguing and adds additional insights into how GLD trading is flowing into and impacting the underlying gold market. Realize that each share of GLD represents 1/10th of an ounce of gold. So if you want to mentally convert GLD volume into effective gold volume traded by American stock accounts, then just divide the left axis by 10.

Whenever an asset grows more popular, which is virtually inevitable the longer a price rises on balance, trading volume increases. We have definitely seen this in GLD. The straight line drawn through GLD volume above is a mathematical best-fit line. So average GLD volume has soared from around 2m shares a day to about 7m now in just three years.

This is all the more remarkable considering gold was consolidating for over half of this period. Generally during consolidations, trading interest fades as traders get bored. Without rising prices, traders usually gradually exit a market and look for greener fields elsewhere. And while we did see this to some extent in GLD, its volume waned a bit during consolidations, it was still rising on balance. This means interest in GLD, the ETF's popularity among stock investors, was steadily growing.

But this raw share-volume growth really understates GLD's trading impact. Capital volume should also be considered. Capital volume is shares traded multiplied by the share price. If you tell me you bought a million shares of a stock yesterday, but it was one trading at just a penny, I won't be too impressed. But if you bought a million shares of Google, holy cow that is a mighty big line you are swinging. Volume must be considered in concert with share price to really see growth trends.

Back in early 2005 when GLD was young and trading 2m shares a day or so, this ETF was only worth about $43. So this works out to $86m in capital volume per day as a rough benchmark. Assuming 7m is now average today, and GLD is trading near $80 per share, we are talking about $560m in capital volume. Thus trading interest in GLD has grown on the order of 7x since its November 2004 debut! Driving this kind of interest among stock traders makes GLD an unqualified runaway success.

GLD's big volume spikes also offer insights into GLD traders' behavior and hence impact on the gold price. Note above that the ETF's biggest volume spikes tended to occur near interim highs leading into their following intense selloffs. Since GLD volume soared on these selloffs, its traders certainly feel fear like all speculators do. But collectively this fear wasn't enough to drive GLD down much faster than gold. This is evidenced by GLD's stable bullion holdings during selloffs even despite these big volume spikes.

It is possible, indeed probable, that in the future some big gold selloff will spook the GLD holders so much that they will dump GLD at a much faster rate than gold. In order to maintain gold tracking in this scenario, the ETF custodians will have to sell off bullion to equalize this sell-side imbalance between GLD and the metal. This will exacerbate and amplify the big gold selloff. But so far at least, GLD's downside impact on gold has been minimal while its upside impact has been substantial.

Now these stock traders are buying GLD because they want exposure to the gold price in their portfolios. So the degree to which GLD actually tracks gold is crucial to this ETF's continuing acceptance and growth. This next chart quantifies the variance between the GLD share price times ten (to equate to one ounce) and a couple of key gold benchmarks. Considering gold's increasing volatility, GLD's custodians have done an outstanding job of maintaining tight GLD tracking of gold.

To evaluate GLD's tracking of gold, its entire mission in life, first check out the yellow GLDx10 line overlaying the blue gold price. As is evident visually, GLD has tracked all the underlying volatility in the metal itself extremely well. Life to date, GLD's correlation r-square with gold is a staggering 99.97%! Thus for all intents and purposes, trading GLD offers the same exposure to the gold price as gold itself.

Despite such a statistical mirroring, GLD's tracking of gold has loosened slightly. While it covered the blue gold line above perfectly in its first year, since then GLD's small but noticeable deviation from the metal is definitely growing. This may bother some traders, but it doesn't bother me a bit. It was not only known in advance when this ETF launched that this was inevitable, but this trivial deviation hasn't materially affected GLD's returns versus gold's. Since GLD's launch day, it is up 85.3% at best compared to gold's 88.9%.

Why is this happening? Like everyone else, the GLD custodians deserve to make a living and earn a profit from their hard work. So like every ETF on the planet, they charge a reasonable management fee to maintain GLD and keep it tracking gold. GLD's is 0.4% per year. If you own this ETF, you will pay 0.4% of its assets annually for this privilege. So after three years, there is about 1.2% less gold per share than there was initially at launch.

This is reflected by the red variance downtrend rendered above. It is the 5-day moving average of the gold close divided by the GLDx10 close. The white line does the same thing with the London PM fix, which is the gold benchmark off of which GLD's custodians choose to measure themselves. GLD's variances on balance are so tight that they only reflect this management fee. GLD, like all other ETFs, is a very-gradually wasting asset. But this is just the nature of the ETF beast, the price for convenience.

The most popular ETF today is the PowerShares QQQ which tracks the NASDAQ 100. As the elite ETF, it is considered the best in the business. The QQQQ custodians charge a 0.2% annual management fee, so each year the value of the QQQQs relative to the NASDAQ 100 shrinks by 0.2%. This is not that much lower than GLD's expense ratio, despite the fact that buying, selling, and moving physical gold is vastly more expensive than instant electronic transfers of stocks. GLD's expense ratio is certainly fair, especially considering the bid-ask and commissions on a single round-trip coin trade can run 3% to 6%.

And with the exception of this necessary cost of running an ETF, GLD has tracked gold nearly perfectly. This is evident both visually and statistically. This kind of flawless record, exposing stock capital to the gold price just as advertised, should work to attract in countless new investors in the years to come. GLD really is the quickest and easiest way for traders to gain gold-price exposure via their usual stock-trading accounts.

Now traditional physical-gold-coin investors still don't like GLD for the most part, and I certainly agree with them that GLD is "paper gold" since it isn't physical in your own immediate possession. If you don't like GLD and would rather own real gold coins, more power to you. I am the same way personally. But despite my own proclivities I am really glad there are alternatives for other investors who don't share my worldview. I won't eat McDonald's hamburgers because I don't care for their taste. Nevertheless, I still think McDonald's is a fantastic company and I am glad it exists for people who do like its hamburgers.

Three years after its birth, traditionalist objections to GLD generally surround two fronts. I would like to address them briefly. The first wonders if GLD really has the physical gold bullion that it claims it has. The second wonders if GLD demand from stock capital is cannibalizing demand for actual gold stocks. If you are worried about either of these factors, rest assured that you aren't GLD's target market anyway.

Does GLD really have all the gold it claims? I don't know. I have never been to GLD's vaults. And if I did go, since I am not a metallurgist I would have no way of knowing if the bars I held were solid gold or gold-plated lead. When you think about it, there is really very little we actually know as investors. I own dozens of mining companies that collectively claim to operate over 100 mines worldwide. But I have never seen one of these mines with my own eyes. Do they really exist? And if I did travel to one, as a non-geologist I'd have no way of knowing if the ore the company showed me really contained the metals claimed or if it was just useless waste rock.

Ultimately, when it comes down to it, all investment relies on trust. We choose to believe what the companies we own tell the SEC, their auditors, the media, and us. And most of the time this system works exceedingly well. For every fraud like Enron, there are hundreds of good honest companies worldwide that are really doing exactly what they are telling investors they are doing. After three years of studying GLD, and looking into all the controversy, I have no reason to believe it is not telling the truth on its gold bullion. I don't know for sure, and I can't know for sure, but this ETF seems perfectly legitimate to me.

On GLD cannibalizing gold-stock investment, I don't believe this theory for a minute. Traders buy gold stocks because they greatly leverage gold's gains. Gold is up 225% at best bull-to-date while the HUI gold-stock index is up 1167%. This is huge 5.2x leverage! Gold-stock traders willingly assume much greater risks plaguing specific companies and projects in order for a shot at far higher returns. Since GLD at best can only mirror gold's gains minus a 0.4% expense ratio annually, it is not a competitor with gold stocks. Gold stocks are a pure gold leverage play, and GLD has zero gold leverage.

Speaking of gold stocks, thanks to their massive gains so far and even greater potential ahead, they are my vehicle of choice to ride this gold bull. Just this week my business partner Scott Wright finished a brand-new report on our favorite 20 gold producers. After starting with well over 100 gold stocks worldwide, we spent four months whittling them down to our 20 favorite producers through deep and extensive fundamental research.

While we do all this hard research work to know what the most promising companies are to recommend to our subscribers and deploy our own capital in, due to popular demand we also sell it. If you are interested, our new report is now on sale. Its new format delves into each company's fundamentals on a project-by-project basis to get you up to speed on fantastic opportunities. If you want to know where to deploy gold-stock capital, and save yourself many hundreds of hours of research, buy this report today.

Back to GLD, the bottom line is its impact in the gold world is already big and is only growing. Whether or not a gold ETF appeals to you, alternative ways to buy gold are very healthy for the entire gold market. The more capital that flows into gold, regardless of source or conduit, the bigger this bull will ultimately grow and the greater the profits will balloon for all of us.

Gold ETFs are a hugely important capital and psychological gateway into gold for non-traditional gold investors. Let's welcome them! The more the merrier. No, GLD is not the same as a gold coin in your fist, but it isn't intended to be. It is a remarkably efficient and low-transaction-cost way for stock traders to gain gold-price exposure via their usual stock-trading accounts. And GLD has excelled in this mission.