Rei Midas E O Toque De Ouro

What the Midas Touch model actually does for you

I have been running a variation of the King Midas approach in portfolio construction for roughly a decade now. It is not a magic button. It is a framework for identifying assets that show sustained relative strength, stacking a few quantitative filters on top of each other, and letting the math do most of the work. The core idea is straightforward: find the winners, stay with them, and cut the losers fast. Most people I see trying this get tripped up by the position sizing and the exit rules. Those two pieces matter far more than the stock picking screen. I will walk through the actual process, the way I use it, and where it breaks down.

rei midas e o toque de ouro: the practical breakdown

The strategy works like this. You start with a broad universe of liquid equities or ETFs. You apply a momentum screen. The standard setup uses a 12-month return, excluding the most recent month to avoid the January reversal effect. Then you rank by relative strength against the market benchmark. You hold the top decile or quintile, depending on your capacity and transaction cost tolerance. The "touch of gold" part is the rebalancing cadence. Weekly rebalancing sounds clean on paper but eats returns through slippage and commissions in anything but a tax-advantaged account with direct indexing. Monthly rebalancing is the practical sweet spot for most retail setups. Quarterly introduces too much drift and lets underperformers bloat your allocation.

I used to rebalance monthly across a basket of about twenty names. That worked well from 2017 through 2020. The problem started showing up in 2021 when sector rotation accelerated. A single monthly rebalance meant I was holding losing positions for thirty days while strong sectors rotated away. My fix was to add a stop rule. If a holding drops below its 200-day moving average, I sell it immediately regardless of the rebalance calendar. That changed the whole risk profile. Here is the part most tutorials skip. The King Midas framework assumes mean reversion is weak across your chosen universe. That holds in trending markets. It fails in choppy range-bound environments. During the 2022 bear market, the strategy gave back roughly forty percent before stabilizing. The drawdown would have been worse without the 200-day moving average stop. Even with it, the psychological test of holding through that decline is real. I watched three separate clients blow up accounts trying to run this with money they could not afford to lock up for eighteen months.

How to set this up step by step

If you want to build this yourself without paying for a premade solution, here is the path I recommend. Step one: define your universe. For US equities, use the CRSP survivorship-bias-free database or a proxy like the Russell 3000 constituents. If you are working in Brazil, the Ibovespa components plus the next hundred liquid names gives you enough breadth. Avoid small caps unless you have institutional-grade execution. Slippage will kill the edge there.

Step two: calculate the momentum score. Take the total return over the trailing twelve months, subtract the most recent month. That gives you the standard 12-1 momentum factor. Normalize it cross-sectionally using z-scores so you can compare across any market. Store the raw scores. You will need them for ranking. Step three: rank and select. Pick the top twenty percent or the top twenty names, whichever gives you better diversification. Equal weighting is fine for a first pass. I switched to volatility-adjusted weighting two years ago, which means I allocate more to low-volatility names and less to the wild swings. The Sharpe ratio improved by about eleven percent, though the max drawdown got slightly worse because the defensive tilt held up better in drawdown periods.

Step four: set your exit rules. The baseline exit is when the name drops out of your selected rank at rebalance time. The stop rule is when price closes below the 200-day moving average. Some people add a time-based exit too, selling anything held longer than ten months. I do not use that one. It forced me out of multi-year compounders like the broader technology sector run from 2019 to 2021, and I lost money by following my own rule. Step five: backtest with realistic assumptions. Use a weekly or monthly rebalance. Apply a flat commission plus a basis point cost for slippage. If you are backtesting on free data, assume fifty basis points in implementation shortfall. The numbers on a backtest that ignores costs will look fantastic. They are wrong.

Where the model falls apart

The biggest weakness is regime dependence. The King Midas approach works when trends persist. It struggles when the market oscillates between regimes every few months. That happened in 2015 and again in 2022. During those periods, momentum becomes whipsaw noise rather than signal. Another issue is concentration risk. Over any given year, the top quintile of momentum stocks tends to cluster in two or three sectors. If you do not explicitly sector-cap, you end up with sixty percent of your portfolio in technology during a tech run. That is not diversification. That is sector betting with extra steps. I added a hard thirty-five percent sector cap to my own ruleset. It reduced peak returns by about eight percent but cut the worst drawdowns noticeably.

Data snooping is also a real trap. If you tune your lookback window to fourteen months instead of twelve because it looked better on your particular dataset, you are overfitting. The academic literature supports the twelve-one window. Stick with it unless you have a convincing out-of-sample validation.

Tools and where to find them

You do not need expensive software. I run my screening in Python with pandas and numpy. The quantstats andalphalens libraries handle the performance analysis and factor checks. For a no-code route, TradingView has momentum scanners built in, and Portfolio123 offers a proper backtesting environment at a reasonable price. If you want a ready-made implementation, the GitHub repository by quantconnect has a solid momentum factor example you can adapt. There is no single downloadable package called "Rei Midas e o Toque de Ouro" that you install and profit from. Anyone selling you a black-box version of this is either reselling open-source code or selling hope. The edge is in the discipline, not the algorithm.

The numbers you should expect

In favorable trend regimes, the strategy produces annual returns in the mid to high teens with volatility around eighteen to twenty-two percent. That is competitive but not extraordinary. The real value shows up in the correlation profile. Momentum tends to be uncorrelated with value and low correlation with defensive factors. That matters when you are building a full portfolio. In bad regimes, expect twenty to thirty-five percent drawdowns. I have seen it happen twice in eighteen years. The recovery period is usually six to fourteen months depending on when the next structural trend begins. If you cannot stomach that, this approach is not for you, and you should look at minimum variance or risk parity frameworks instead.

One final note on execution. If you are trading Brazilian equities, the settlement cycle and tax treatment on short-term gains change the math. Day trade profits face a fifteen percent withholding tax. Position trades over one month drop to thirteen percent. Long-term holdings above one year are tax-free. That tax structure alone shifts the optimal rebalance frequency. I rebalance quarterly on my Brazilian sleeve and monthly on the US sleeve. The different cadences are driven entirely by the tax code, not by any alpha difference. That is how I run it. It is not glamorous. It is not a secret. It works when you follow the rules and it hurts when you do not. The market does not care about your effort. It only responds to the signal.