White Paper
Project History & Objective.
The EPIC software development team set out nine years ago to develop a sophisticated software architecture for trading crude oil futures. Subsequently, other market instruments of trade were added including Equity Market Indices, Stocks, Commodities and Crypto.
The objective: to provide a stable, high performance auto trading system with continuously increasing ROI over time as the algorithm processes market data, structure of trade action and order-flow.
The software has undergone a number of updates over the years. It is now a stable trading entity that continues to excel in terms of ROI.
Current Deployment and Use.
EPIC is currently used by a diverse number of trading entities. These include the trading team at Compound Trading Group and associated trading platforms (for live trading alerts), auto-trader platforms, automatic copy trading, private accredited investor account trading and our five tokenized crypto projects.
As of late 2024 the software is considered “core” architectually complete and only minor maintenance and incremental updates are expected going forward.
Near Term Plans.
The EPIC Development Team is now in the process of deploying EPIC Machine Learning Trade Software to a Private Equity Platform for the purpose of securing a quantifiable data-set in early 2025. This will be used as official data for deploying EPIC Software in additional Private Equity Market Organizations & Funds and various Professional Trading Groups. We are also targeting late 2025 or early 2026 as a time to enter the publicly traded markets, pending private equity deployment.
2025 is the year EPIC Machine Learning Software is deployed through-out global markets via Private Equity Fund and Professional Trading Groups.
Performance (ROI) and Stability.
EPIC is now deployed in 27 different instruments of trade on various trade platforms and exchanges, with many different account sizes.
The team of five computer scientists running the operation are housed in a secure facility with 27 servers (each running one instrument of trade) and various other network equipment. The total upstart development cost (including overhead) to date has been 9M USD as of late 2024.
Annual returns range between 30% ROI – 300% with the average return being just over 80% ROI per annum. There are few real world trade scenarios in which EPIC is returning below 30% and few over 300% per year. The returns are expected to escalate at a pace of about 10-30% per year as it becomes more refined over time. Under most market conditions a 70-90% per year ROI is the base-line expectation for new accounts.
It is important to note that the specific instrument of trade, size of the account and when in the trade cycle an account is onboarded to the software will affect the ROI. For example, an account onboarded mid cycle of trade sequence can see varied results until the onboarded account has run a full trade sequence. This can be a few months or more depending on the market conditions and instrument of trade.
The two primary developments that elevate EPIC above the competition are the EPIC IDENT™ Order Flow code and the Intra-Day High Frequency code that relies heavily on IDENT™.
There are three primary protocols of code in the software;
- Swing Trading (average sequence duration is 11 weeks and has been as short as 1 week and as long as 16 weeks).
- Intra-Day High Frequency (average sequence duration is considerably less than 24 hours but can also be described at times as intra-week lasting up to 3 or 4 days).
- EPIC IDENT™ Order Flow (extremely high frequency for positioning alongside major machine entities that have presented themselves in the intra-day order flow).
The software has proven to be very stable while consistently providing considerable returns.
To our knowledge of publicly available nformation – the EPIC Machine Trading Software is best in class worldwide.
What Do Best in Class Returns Look Like?
Jim Simons achieved an astounding average annual return of 66% from 1988 to 2018, before management fees. Contrast this with Peter Lynch, 29%, Warren Buffet, 21%, George Soros, 20%, Charlie Munger, 20%, and with the average annual return on the Standard and Poor’s index of 10%.
From Google AI:
“Jim Simons, a legendary investor and mathematician, achieved an average annual return of 66% for his Medallion Fund over 40 years. This return rate is considered unrivaled in the investment world.[Text Wrapping Break]Here are some other details about Jim Simons’ returns:[Text Wrapping Break]Net returns: Simons’ average annual net return was 39.2% after fees.[Text Wrapping Break]Fees: Simons’ fund charged a 5% fixed fee and a 44% performance fee after 2002.[Text Wrapping Break]Comparison to the S&P 500: In 2008, the Medallion Fund made 82% net of fees, while the S&P 500 lost 37%.[Text Wrapping Break]Comparison to Warren Buffett: While Simons is considered by some to be one of the most successful investors in history, his investment strategy was very different from Warren Buffett’s.[Text Wrapping Break]Simons was known as the “quant king” for his pioneering use of quantitative analysis in investing. He founded Renaissance Technologies in 1988, convinced that he could predict market movements with the same certainty as mathematical equations.”
References:
https://www.voronoiapp.com/markets/Jim-Simons-Medallion-Fund-vs-SP-500-Returns-1558
The main takeaway, EPIC Machine Learning Software is outperforming the best in the world, and it is doing that with superior stability (drawdown relative to return).
EPIC Software Highlights.
- Lightning Fast Decisions. EPIC trading software executes trades through utilizing over 9300 weighted decisions instantaneously. The instructions provided within the architecture continue to grow. A human trader simply cannot process this amount of data and make decisions as quickly or precisely.
- Algorithmic Chart Models. The EPIC software code includes over three hundred proprietary algorithmic chart models and the catalogue is growing. The algorithmic models have been designed, tested and refined in real-world trade for over 9 years by a team of experienced day traders. The trading models represent all time-frames from 15 second to monthly time-frames of trade. The algorithmic models have been back-tested to sixty months historically.
- Conventional Charting. The software includes conventional charting structures on all time-frames, also back-tested sixty months.
- Common Trade Set-Ups. The software includes common trade set-ups that traders routinely implement.
- Order Flow. EPIC IDENT™ is a data-driven order flow intelligence system that makes real time decisions. The software consists of a proprietary order flow identification system that tracks behaviour of other large machine trading entities and weighs their historical trade patterns in its decision tree. EPIC IDENT™ increases its intelligence as it gathers data intra-day specific to liquidity flow, historical patterns, time of day, volatility, various preferences, latency, rejects and more. The method is similar to back-testing charting. However, the process occurs in real-time. In short, the software is looking for “fingerprints” within market liquidity. We cannot back-test 60 months as with charting, but back-testing from the start of software deployment has been achieved.
- Time Cycles. Time cycles are within all algorithmic and conventional trading model structures. Order flow also has identified time cycles and other time cycle events such as, for example, weekly reporting in oil markets (API, EIA and rig counts). Additionally there are time-of day market time cycles around the world. All of these different time-cycles are included in the software architecture.
- Hard-Pivot Architecture. The risk threshold – management system within the EPIC architecture has a hard pivot rule-set that has near ended substantial risk to accounts.
Combined, these advantages enable the EPIC Trading software to consistently outperform conventional trading methods.
About Public Market Machine Learning Trade.
The world of public market trade is rapidly changing. It is estimated (depending on source) that over 80% of futures are now traded by machines, not humans.
Machine trade may be simple, bot style software, high-frequency software or more sophisticated architecture as with EPIC.
Our team commenced the trading software development journey nine years ago with algorithmic chart modeling. From day one we employed computer scientists to work with us on a daily basis to build software that would emulate our trading methods.
Over time the software started to win more trades than our traders and today we rely almost solely on the software to execute trades. We simply “tweak” the software at each trade sequence to improve performance.
Account Size: ROI and Draw-Down Volatility.
The larger the account traded, the easier it is for the software to limit downside risk and provide optimal returns.
We have learned over time that the smaller the account traded size the more ROI instability and associated risk. For example, with crude oil futures trade, a $600k account becomes a stable size and will rarely see ROI volatility and draw-down risk outside of expectations. A $900k account size or larger will never encounter considerable ROI volatility. As noted above, the risk threshold management system within the EPIC architecture now uses a hard pivot rule-set that has nearly ended the risk of major drawdown for larger accounts.
The software is designed to trade a given sequence within structures or set-ups. As the market price changes, the software trading logic uses all available data to update its decision tree.
You can imagine this as a dot plot process similar to the game “Go”. This helps to visualize how the software plots a trade sequence.
The regular ups and downs of price movement allow the software to plot a plan of trade within a sequence. The larger the account, the more dots that can be plotted. Trades are positioned with these ‘dots’ along with the ebb and flow of the markets.
To understand the full trading methodology of the software in each sequence of trade requires a moderately in-depth conversation to review data with a member of our team.
In short, the software uses a positioning (swing trade) protocol for each sequence of trade and a high-frequency intra-day protocol via EPIC IDENT™ technology. On the positioning (swing trade) side of the architecture, this means that as price is rising the software is building a position short through-out the rally (and the opposite is true if the price is falling). However, the average cost is off-set by the high-frequency component of trade via EPIC IDENT™ technology so that when the trend reverses the software achieves considerable returns.
API / Deployment Architecture.
EPIC software is designed to be deployed remotely – accessing an account and executing trades. This provides the account holder with ultimate control. The account holder grants the software access and the software executes machine trades to the account. This architecture provides the opportunity for decentralized platform integration.
Conclusion.
This paper outlines the opportunities that can be presented by the growing influence of machine trade on global financial markets.
Competitors within the machine trade industry are becoming more and more refined and successful. The best in class are assumed to be winning a progressively larger proportion of the earnings.
The most significant immediate challenge developers face in machine trade is building a product that will perform within a threshold of downside limiting stability while outperforming conventional trading methods.
Thereafter the challenge becomes competing in ROI against other machine trade peers.
It is our expectation that fewer and fewer competitors will achieve more of the proceeds in public markets at an exponential rate. This creates an element of urgency in development and deployment.
The EPIC machine trading software achieves consistent, predictable and adaptable architecture to provide best in class returns.