-1 Modeling

Modeling "after -1" argues that ShezheaNET MLT_HFT outcome (A) is always false. Modeling "after -1" allows for reverse beneficial context modeling, meaning everything is considered false, even when the outcome is correct. We found a stochastically advantageous development considering the -1 model subject. If (A) is always false, meaning the Adjusted Parameter Demo Test will always fail it creates a feedback loop in the negative rewarding system, trim-tuning the parameter models. (A) as an outcome is always false, until (B), the 20x3 outcome is equal to (A). A having the same int as B, consider it here being the parameters (q/a/y/r/d), means there is a perfect probability of executing the past trading environment (TE).

$$\text{Maximize } TE = f(q, a, y, r, d)$$

Finding (q/a/y/r/d) allows for categorizing and finding patterns in different TEs, resulting in the possibility of specific training of submodels on top of the main MLT model. 1/9 TEs are categorized as the following in NQ:

• Optimal 1/9 NQ Model (Sub-Model A=) -> oNQ9

• Less optimal 1/9 NQ Model (Sub-Model A&) -> loNQ9

• Zero optimal 1/9 NQ Model (Sub-Model A/) -> zoNQ9

• Optimal 1/9 SeekDestroy NQ Model (Sub-Model A$) -> skNQ9 Multisided Fraction Executions in skNQx Preview: skHFT - Liquidity Flow for markets

• (Broken 1/9 SeekDestroy NQ Model (Sub-Model A§) -> bskNQ9)

1/9 in this context means 1/9th of a second. We are releasing the detailed names of the models, to avoid misunderstandings in future press releases.

oNQ9 MLT-HFT Backtesting via the -1 Method

Execution 1 with the outcome (A1) is false (/ has a negative result), resulting in the (B1) data-cluster with 20x3 outcome being (naturally because of -1) false as well, since (A) ≠ (B). This allows for an optimized finding of execution parameters (q/a/y/r/d) based on ticker parameters that were present before Execution 1. Note that the algorithm doesn't act according to the target to replicate (q/a/y/r/d) to the following Executions, but to categorize it in TEs. Here Execution 1 gets categorized as an Optimal 1/9 NQ Model, after finding the exact optimal parameters for Execution 1. TEs are nothing less than periods (usually less than 20 seconds). The MLT-HFT algorithm can now adapt to different TEs at a lightning-fast speed. This fixes HFT overfitting. The following concludes the basic framework behind this.

# Define the main ShezheaNET MLT_HFT model
struct ShezheaNET
    parameters::Dict{Symbol, Any}
    submodels::Dict{Symbol, Any}
end

# Initialize the model with default parameters
function init_model()
    parameters = Dict(:q => 0, :a => 0, :y => 0, :r => 0, :d => 0)
    submodels = Dict()
    return ShezheaNET(parameters, submodels)
end

# Function to determine if outcome A is always false
function is_outcome_A_false(model::ShezheaNET)
    # Logic to determine if outcome A is false
    # This is a placeholder for the actual logic
    return true
end

# Function to handle the -1 model and its implications
function handle_minus_one_model(model::ShezheaNET)
    if is_outcome_A_false(model)
        # Adjust parameters for feedback loop in negative rewarding system
        # Placeholder logic for trim-tuning the parameter models
        model.parameters[:q] -= 1
        model.parameters[:a] -= 1
        model.parameters[:y] -= 1
        model.parameters[:r] -= 1
        model.parameters[:d] -= 1
    end
end

# Define sub-models based on TE categorization
function define_submodels(model::ShezheaNET)
    model.submodels[:oNQ9] = "Optimal 1/9 NQ Model"
    model.submodels[:loNQ9] = "Less optimal 1/9 NQ Model"
    model.submodels[:zoNQ9] = "Zero optimal 1/9 NQ Model"
    model.submodels[:skNQ9] = "Optimal 1/9 SeekDestroy NQ Model"
    model.submodels[:bskNQ9] = "Broken 1/9 SeekDestroy NQ Model"
end

# Main execution
model = init_model()
handle_minus_one_model(model)
define_submodels(model)

# Print model status
println("Model parameters: ", model.parameters)
println("Model submodels: ", model.submodels)

This is just the framework, not a functional code.