Trader with 30+ Years of Experience Reveals His Strategy Secrets

In this interview, Christos—an experienced trader active since 1987—shares his full approach to building and running live algorithmic strategies on Alpaca using StrategyQuantX and AlgoCloud.

He talks openly about:

• The 7 strategies he runs live (and why they’re long-only)

• How market regimes determine when his systems trade

• His method for robustness testing and avoiding overfitting

• Why strategy simplicity often outperforms complexity

• The tools and resources that shaped his success

Christos doesn’t just talk theory—he shares real setups, performance insights, and decision-making processes from decades of experience. Whether you’re developing your first strategy or managing a portfolio, this interview offers serious, practical value.

🎥 Watch the full conversation now on YouTube and learn from someone who’s built profitable systems that last.

Transcript:

I’ve been a trader since 1987.
Today I have a total of seven strategies running on Alpaca Live.
Last Wednesday, a week ago, some of the algos made the most money they ever made.
And I’d like to point out a very valuable feature of AlgoCloud and SQX.
I think the most valuable thing for traders to pay attention to is the market regime.
I like the robustness testing.
That’s a very critical part of the overall process.
Hello, everyone.
Hello.
In our podcast about algo trading, which is part of the current StockLab course,
which is ongoing.
And it’s my pleasure to introduce today Christos,
who is a professional trader for many years,
has a very interesting path of a trader and he’s a successful trader.
So there will be a lot of insights.
For example, right now, as we are recording the interview,
there is high volatility in the market due to the situation in the United States.
And Mr. Trump steps and Christos will touch how he’s dealing with it
and basically how he’s successfully dealing.
So again, hello, everyone.
Christos, thank you for joining us today.
And please tell us something about you.
You’re very welcome, Cornel.
It’s a pleasure of mine to meet you online and discuss some of the
approaches and successes I had with AlgoCloud.
I’ve been a trader since 1987.
I started my trading journey just doing simple stock trades back then.
And then I transitioned to full-time trading in 2000.
In 2000, I started exploring some automation for executing trades.
But I concentrated mostly on index products, such as the S&P 500 via the SPY ETF.
And later, I transitioned to the ES futures.
But I was always interested in doing stock trading automatically.
But I never seemed to find the proper tools, the proper software tools
to, first of all, do the data exploration, the data mining.
And then to transition whatever strategies I was coming up with onto a specific platform.
About 14 months ago, I came across SQX and AlgoCloud.
And I immediately realized this is a very powerful tool.
And it was what I was looking for.
I like the fact that the tool is highly integrated.
I like the fact that there’s a data
subsection in the tool that handles the data, that you can quickly formulate a strategy
with some simple code.
And also, I love the fact that it executes the backtesting with amazing speed.
After coming across the tool 14 months ago, I started implementing a number of strategies,
which I thought would be successful given my experience from the discretionary part of my
trading. And today, I have a total of seven strategies running on Alpaca Live.
And I will go now into details of what I have on Alpaca Live running
to give a sense of the blend, of the combination of the strategies that I have,
which incidentally have low correlation among each other.
I think they are less than 0.4 correlation with each other.
I will start with the first strategy that has helped me in this high volatility environment.
It’s a gap strategy based on the Russell 3000.
And basically, it looks for volatile conditions in the market and specifically when the market
opens with a specific gap. And then from that, it has a special
ranking function, which determines which of the universe of the 3000 stocks in the Russell
3000 are the most promising to enter a long position. Incidentally, all my strategies
are long only strategies. So this particular one, the gap Russell 3000, has been very
valuable in this high volatility environment because for the last month, the market tends
to open, the real-time trading hours tend to open with a gap. In addition to this particular
strategy, I have a seasonal strategy, which operates on the S&P 500. And I like to point
out a very valuable feature of AlgoCloud and SQX, the fact that you create your own
stop group. So what I have done is I’ve looked for strong seasonal patterns at specific dates,
let’s say from the middle of one month to the middle of the next month.
And then you can create a universe that just contains those names. And then you can apply
a mean reversion or a trend following or breakout strategy on this refined universe of stocks.
This has been a very successful approach. And I think it’s valuable the fact that
SQX and AlgoCloud gives you that capability. So instead of operating on the Russell 3000,
NASDAQ 100, S&P 500 universe, you can actually define your own universe that the strategy will
operate in. Of course, with the ranking function and so forth, but it’s an approach that I found
to be valuable. I have another approach, which is a combination of momentum and seasonal
and operation on the S&P 500. And this particular strategy also takes advantage of the fact that in
the AlgoCloud, you can define secondary time series. So when you’re doing stock picking,
you can define SPY as your second time series. And if the SPY itself
behaves in a particular way, then it allows the strategy to enter different positions.
By the way, I think the most valuable thing for traders to pay attention to
is the market regime, whether we’re in a trending up market, trending down market,
we’re in a low volatility environment or high volatility environment.
In fact, quite a few of my strategies shut down in a high volatility environment.
Now, how do I measure volatility? Alpaca doesn’t give me the option of accessing the VIX.
So what I’ve done is I use the function within the AlgoCloud.
So the function is called variance. And what I do is I compare the variance historically.
So I look how the variance is behaving presently versus the past. And if certain conditions are
met, some of the strategies completely shut down. Another strategy I have looks at the breakout
universe in the Russell 3000 collection of stocks. So this one is strictly a breakout strategy that
that attempts to reduce the drawdown. Drawdown is always a concern.
Return to drawdown is always desirable to be as high as possible.
Another strategy I have is a mean reversion that operates on Nasdaq 100. I noticed the Nasdaq 100
stocks tend to be strongly mean reverting, potentially because a lot of money managers
are looking for opportunities to get into one of those Nasdaq 100 stocks. But also this particular
strategy is paying attention to what’s happening in the overall market. We don’t want to enter when
obviously the market is in the freefall. So I’m looking also what’s going on with the SPY ETF.
Another strategy I have is a continuation strategy in the Russell 3000. This is a very short-term
oriented strategy. It takes position after it sees a small return to particular value
and then with the expectation that the next day will see some corrective action to the upside.
So it’s a strategy that tries to capitalize on momentary backtracking of a stock price.
And finally I have a very classic mean reversion strategy that operates on the S&P 500.
And there I use again the market regime filter. So all in all I have
presently seven live strategies running on Alpaca Live with quite good results.
Some of them have completely shut down. They’re not taking positions anymore. In addition to these
seven live strategies on Alpaca Live, I have 59 strategies running on Alpaca Paper.
These now are going through an evaluation phase. And from those 59 strategies, the five are
operating in stock picker mode and 54 they operate in the single asset mode. And one of the reasons
I’m doing the single asset is because there is a strong interest from an institutional client to
evaluate how these strategies perform on specific assets. Those specific assets
are selected based on macro themes. So that’s how they operate and I’m trying to
accommodate them using the single asset
approach. Now, would you like to ask me anything? I don’t want to monopolize.
I don’t want to interrupt you because you are sharing such valuable insight. I didn’t expect
that you’ll go deeply and openly that way. I’d like to discuss what you are actually running
because many times it’s just secret. It’s like secret sauce. I know.
And so I don’t want to interrupt you and just continue. I think sometimes traders are overly
protective. The big challenge is how much capital to allocate
and how to control risk based on the assets volatility. I think that’s where the real
secret is and how to combine all these strategies in a portfolio. That’s where I
think the essence that the challenge is. How to combine the strategy in the portfolio and what
capital to allocate on a per strategy basis. More or less people know how to do a min reversion or
or a breakout strategy. The information is out there. In fact, I will mention the books that
help me. The sources that help me do some of these strategies. In fact, I’ll do it right now.
I have to say that I got a lot of inspiration from Malik Keis’ YouTube channel. I think he’s
putting out a valuable content and if people pay attention to the content they can come up
and using of course the algoplot and the sqx tool they definitely can come up with money making
strategies. But in addition to inspiration from Ali Casey’ YouTube channel, I have an extensive
book library and over the years I keep referring to this library and in fact those books are
valuable in the sense that were published a long time ago and a lot of the strategies effectively
are out of sample right now and it’s very instructive to see how these strategies are
performing using of course a state-of-the-art tool such as sqx. May I ask you, can you
recommend for example some book? I have them right here. I have a book called
Trade Like a Hedge Fund by James Altucher. It’s a small book. I don’t know if it’s out of
print but I think people can easily find it on the internet. The other book I like very much is
In the Trading Cockpit by Gil Morales. Gil Morales was the right hand of O’Neill, who was a famous
investor and money manager. Another book is the Alphan book also by Gil Morales
and of course the all-time classic, The Long-Term Secrets into Short-Term Trading by Larry Williams.
Larry Williams is a very famous commodities trader and he has published many books and he
also has a lot of YouTube videos. I think if people go through these sources for ideas
and mix their own preferences for
risk, tolerance and also from their own trading experience, they definitely can come up with
successful strategies. What I like is that within SQX and AlgoCloud you can
very quickly prototype a strategy. I have to give you an idea. My Gab Russell 3000 strategy
has three lines of code. It’s just three lines of code. People don’t realize how useful the
ranking function is. The ranking function goes through three thousand stocks and of course
if your conditions are met, the ranking function has to do the last filtering.
But essentially the combination of the code in the main body of the logic and then
with the addition of the ranking function, it’s a very powerful combination.
So this covers my sources for ideas.
Now I’ve put together here what I like about sqx and AlgoCloud.
I like the fact that it’s a one-step solution. You don’t have to run around and use one software
package to do the downloading, another software package to run the backtesting,
yet another one to do the handshake between the broker and your code
and so forth. So it’s a one-stop solution.
It’s how easy you can code an idea. You don’t need extravagant programming capabilities.
You can express an idea very quickly and I think
I found most of the functions that I was looking for embedded in the sqx. In fact,
I haven’t had difficulty with the functions provided. I’ve done all the work with the
functions provided within the sqx and the AlgoCloud. I like the robustness testing.
That’s a very critical part of the overall process. You want to see if your strategy falls apart,
if some of the conditions are perturbed in a specific way. You want to
see if you operate in a stable area of your input parameters.
I’m always amazed how fast the backtesting is. I’m truly amazed by that.
When we started the stock picking, it was like maybe 10 minutes, you know, the backtest it took.
But then programmers did many optimizations and finally they did it in a way that
it’s quite fast. Please continue and maybe later we can discuss a bit more about
your approach and your view on robustness. In fact, the fast backtesting
was a critical factor in using this data mining software.
Back in the old days, it used to take hours and hours. I would put something to be backtested and
go out shopping or whatever for three hours and then come back and still wonder if it’s still
doing what it’s supposed to be doing and so forth. I like the portfolio part of the software
package where you can put all your strategies together and then
see how the overall portfolio is performing. That’s very valuable information there.
The ease of placing strategies is unbelievable to me that you can have an Alpaca account,
use the AlgoCloud editor and then just push a button and follow a very quick process and
there you are. You’re there. I was amazed the first time it happened and I, you know,
when I got the first reports that the trades were going through,
I was very happy about that.
And that’s true not only for the live but also for the paper accounts.
And I like also the way now AlgoCloud reports on the results, the performance results.
If you have like a number of strategies, you can get the report on a per strategy basis or you can
get the overall report on a portfolio basis. And, you know, I don’t know how would people do it
otherwise, you know, downloading the data in some Excel spreadsheet and then, you know,
fight with Excel and so forth. I think that’s a very valuable tool to have,
the immediate reporting of your equity curve. And of course, I mentioned the construction of
custom stock groups. You can do custom stock groups in SQX and they can do the same on AlgoCloud.
So that’s basically the overall approach that I’m using and where I stand in terms of the
strategies running now on Alpaca.
Yeah. Thank you. Thank you for all the knowledge and also sharing the points you mentioned.
You mentioned what you like. Also, it’s worth to say that you are a long time with us and you are
already kind of part of our team because you helped us to do a lot of
work to handle all the unexpected issues correctly and etc. So it’s finally easy,
but there is a lot of hidden behind and luckily we’ve been able to go through this way and now
we are sure that it’s in the phase that basically it can be used. Many complex things can be used
by just pushing the button and we are glad that we were able to finally reach this phase.
And may I ask you, Christos, you mentioned robustness testing that you like and also
you mentioned the advocacy videos. And may I ask you maybe just in a few points, can you tell
how you look at strategy? If it’s robust, what is sign of robustness for a strategy?
And maybe when a strategy is stopping working, do you have any checklist of any process,
how to evaluate that strategy simply should not be run anymore?
Yeah. Basically, I do the Monte Carlo simulations. Before I even entered the Monte Carlo simulations,
I actually manually vary the parameters to get an initial feel. Because a strategy that
does not work, it actually collapses with a small perturbation of the input values.
So before I enter the Monte Carlo simulations, I do the manual checking. Now if it passes through
the manual checking, I go through the process that is provided on the SQX and I look for a
return to drawdown, I look for the profit factor, profitability, all these parameters. I’m looking
to see how vulnerable they are to the perturbation of the input values. I find particularly useful
the 3D mapping to give you a very quick look how stable your parameters are, how stable the
strategy is. Now most of the strategies I use, I would say 95% of them are very simple strategies.
Therefore, I don’t have a multi-parameter state space in order to avoid overfitting. So
the process of checking for robustness is fairly quick because I don’t have a multitude of input
parameters to vary. So basically that’s what I do. There is a manual phase and then there is
the utilizing the tools embedded robustness testing. Now I’ve done a lot of robustness
testing also on single asset strategies. These single asset strategies, if the assets are mildly
correlated, they should show the same behavior. So it’s an indirect way to determine the robustness.
For example, if you have an asset which belongs in the oil exploration field
and you have a strategy that shows profitability on Exxon Mobil
but does not show profitability on some other petroleum company, then there’s obviously
something wrong with the strategy. Also the strategy for Exxon Mobil,
if some variation of the parameters causes maybe another petroleum company to become
profitable, then that’s also a red flag. So I’m looking in different ways to see
how robust the strategies are. So that’s basically my approach when it comes to robustness.
I understand. And how do you deal with the turning off strategies? Because some of them
maybe run forever, but obviously it might happen that strategies stop work.
And what is your view on this topic or your experience?
I first test all the strategies with a 25-year horizon. I want to see how well
they’re performing throughout different market regimes. So basically,
of course, I do some walk-forward testing and out-of-sample testing. But essentially,
I don’t break down ahead of testing the periods, the time periods. Instead, I’m relying mostly on
secondary time series I talked about before, whereby every strategy that picks from the
Russell 3000 or the S&P 500 universe or the NASDAQ 100 universe is looking simultaneously
at what’s happening in the overall market. If it’s a NASDAQ 100 strategy, I’m looking
at what’s happening with the QQQ and also the volatility of the QQQ. If it’s
dealing with the S&P 500 universe, I look at what’s happening with the trend and
the volatility of the SPY ETF. So basically, I don’t aspire too much into going and selectively
turning on and off strategies, depending upon whether they show profitability or breaking apart
easily. If I see some strategy giving me, you know, from backtesting, I have some strategies
that may give me two consecutive months of losses in the backtesting. Well, if I see the same
strategy showing four months of losses under live conditions, then obviously something is
happening there. But I’m very reluctant to turn off a particular strategy just because
it’s showing a couple months of losses. So that’s what I’m doing. I’m relying mostly on having a
secondary time series based on the SPY or the QQQ and determining from there what kind of
market regime we’re in. And I’m relying on that to either enable or disable the execution of the
strategy. Thank you. Thank you for your explanation. It’s very interesting that
you are very stick to and checking this market, those indexes and regimes.
You mentioned many times during the interview that basically, first is regime and then if
regime is fine, then the strategy is trading. It’s important because many traders just develop
the strategy and maybe added some trend filter, but really not care too much about the regime.
And as you are pointing, it’s very important to simply to let strategy to operate in the
friendly or environment which the strategy was created for. Right, exactly. Great.
Yeah, let’s continue.
I’m exploring now a set of strategies that are operating only on an extreme low volatility
environment and they show a lot of promise. So those strategies are inappropriate for the
current environment. But one potential area for someone to look into is, let’s say out of the
Russell 3000 universe, to actually focus on stocks that have extremely low volatility.
Because the results, they may not be spectacular in terms of profitability and in terms of
return on capital, but they’re very stable results. You can come up with a
large sharp ratio and also a large
profit to drawdown ratio. So I think it’s important to focus on this regime
aspect of designing your own algorithm. Thank you. And also, as you are speaking about regime,
when you compose portfolio, you are running like seven algos, was it composed to cover
different phases of the portfolio composition was to cover different phases. So basically,
it kind of simply if the strategy operate, let’s simplify it like closes below and below or up
or up to 200 moving every 200. So let’s use this simple filter closes below or up. And
then in portfolio, you choose two strategies, basically, and one would operate if the
condition is true and second if false. And this is the way we are looking on portfolio creation.
Yes, because your profit targets and stop loss and so forth are greatly determined by the volatility
and you have basically the following challenge either not to be active in a very volatile
environment and or be active in a high volatility environment knowing ahead that
it’s a riskier environment, but also the opportunities are greater.
For example, this past Wednesday, a week ago, some of the algos made the most money they ever made
because the market went up 10%. But they had to withstand the deep dive of the market. So,
you know, those were taken into consideration that they’re operating in a high volatility
environment. So the allocated capital was smaller and provision for the stop loss, the stop
loss setting was very generous to be able to handle this type of market. But as I said before,
most of the strategies deploy this filtering within themselves, but some strategies also they have
their provision for a high volatility environment. So basically, if a developer or
trader approaches the market as a risky affair, so to speak, you know, it focuses first on the risk
and then on the profit. They’ll be ahead of the curve.
Yeah, I understand. Yeah, so I’m glad that maybe, or let’s say if you would like to share anything
more with the traders, then it’s the right time now. Or if not, then I think we covered
kind of a whole process from the creating strategies to gathering some know-how,
composing portfolio, and also tools you are using and robustness approach. So I think it’s like a
whole package which trader needs to start and can get a lot of inspiration from it.
But if you like to share anything more with the traders, let’s tell it.
I’d like to share also the psychological aspect of this.
Okay.
Of developing algorithms. It’s easy to get discouraged in the beginning.
I have to, because of my electrical engineering background, one of the most valuable metrics in
communication is called signal-to-noise ratio. That is how prevailing is the signal versus how
much noise there is in the channel. And markets are incredibly noisy, incredibly chaotic. So it’s
easy for someone started exploring the developing algorithms, it’s easy to get discouraged.
But it’s something that can be
ameliorated by looking at sources for trading ideas. And basically,
they can start putting together some very simple ideas. As I mentioned, there are some sources that
can get some initial ideas. And they can do some modeling with SQX and Alveocloud and see where
that gets them. And the challenge is there, if you come up with a simple strategy, is that
occasionally the market will go into this, if you’re dealing with a long-only strategy,
the challenge is to
take care of the times when the market takes a deep dive. And that’s what
throws most of the strategies off. So there are two ways to handle that. Either
you actually put switches in your strategy that will completely shut off,
or you include in your portfolio some reverse ETF, like
QQQ, for example. You make it that part of your portfolio. And when the markets go in a deep
drawdown, your portfolio, if it deploys some hedging mechanisms, such as the QQQ,
will even out these points in time where the equity curve takes a dive.
Of course, that comes at the expense of the profitability. So that’s something that you need
to design up front. That’s why I said before that coming up with how you combine all these
strategies into a portfolio, that’s the real challenge. So one of the questions is, do you
include some strategies that actually make money when the market goes down? The answer is yes,
you should include some strategies that make money when the market’s going down.
Markets have a long-term upward trend. There’s an upward bias. But it’s worth to consider
some strategies that also make money in a downturn.
I understand. Yeah, thank you. Thank you for this note, for this point. Because
yeah, there are periods when it’s easy to make money. But obviously, there are periods like now,
and then you really see what you have in portfolio and if it can survive.
There is a third way, Cornel. The third way is to actually trade assets that are not
highly correlated. You know, you can have gold and oil and silver and other
ETFs, for example, that deal with this asset classes, which are also part of the alpaca
universe. And I’m actually exploring now this into having a portfolio of just ETFs
that includes both long only ETFs and short only ETFs and blend them together
to even out the equity curve. This is part of my ongoing investigation right now.
Because it’s so easy. Within the alpaca platform, you can just call out the
appropriate ETF and you don’t have to run around and go to different platforms. It’s all there.
It’s all there. It’s just a matter of what assets you choose, non-correlated assets,
also come up with a collection of non-correlated strategies on a per asset basis,
and then combine all these strategies in such a way and allocate the appropriate amount of capital
that will yield the optimum equity curve. It’s a long process. It’s not an easy process. It takes
time. But like I said, SQX quickly can give you results. And that’s a big plus.
Big plus. Yeah, it’s great. You also shared some current work and current
way of thinking and what you are exploring. And it can be also inspired for the listener of this
podcast, what areas they can explore too. And maybe they will share with us some
interesting points and can get inspired by each other. So I think it was a great time,
Christos, and thank you very much. Likewise. I enjoyed sharing all this knowledge. You were
well-prepared for the interview and covered it like a package. I believe that every trader who
is starting and considering should watch it and get a lot of inspiration on any part of the way
he is already, right now, and get something for him. So thank you very much. You’re very welcome.
Very welcome. Always a pleasure to see you. Bye.