little bit of progress

src/fit.jl

@@ -130,6 +130,7 @@ __default_optic(m::MyModel) = OptArgs([
 
 """
 __default_optic(m::Yield.Constant) = OptArgs(@optic(_.rate.value) => -1.0 .. 1.0)
+__default_optic(m::Yield.MonotoneConvex) = OptArgs(@optic(_.rates[*]) => -1.0 .. 1.0)
 __default_optic(m::Yield.IntermediateYieldCurve) = OptArgs(@optic(_.ys[end]) => 0.0 .. 1.0)
 __default_optic(m::Yield.NelsonSiegel) = OptArgs([
     @optic(_.τ₁) => 0.0 .. 100.0
@@ -222,7 +223,7 @@ The default solver is `ECA()` from Metahueristics.jl. This is a stochastic globa
 ## Defining the variables
 
 An arbitrarily complex model may be the object we intend to fit - how does `fit` know what free variables are able to be solved for within the given model?
-`variables` is a singlular or vector optic argument. What does this mean?
+`variables` is a singular or vector optic argument. What does this mean?
 - An optic (or "lens") is a way to define an accessor to a given object. Example:
 
 ```julia-repl
@@ -237,7 +238,7 @@ julia> lens(obj)
 "AA"
 ```
 An optic argument is a singular or vector of lenses with an optional range of acceptable parameters. For example, we might have a model as follows where we want 
-`fit` to optize parameters `a` and `b`:
+`fit` to optimize parameters `a` and `b`:
 
 ```julia
 struct MyModel <:FinanceModels.AbstractModel
@@ -252,7 +253,7 @@ __default_optic(m::MyModel) = OptArgs([
 ```
 In this way, fit know which arbitrary parameters in a given object may be modified. Technically, we are not modifying the immutable `MyModel`, but instead efficiently creating a new instance. This is enabled by [AccessibleOptimization.jl](https://gitlab.com/aplavin/AccessibleOptimization.jl).
 
-Note that not all opitmization algorithms want a bounded interval. In that case, simply leave off the paired range. The prior example would then become:
+Note that not all optimization algorithms want a bounded interval. In that case, simply leave off the paired range. The prior example would then become:
 
 ```julia
 __default_optic(m::MyModel) = OptArgs([
@@ -265,7 +266,7 @@ __default_optic(m::MyModel) = OptArgs([
 
 ## Additional Examples
 
-See the tutorials in the package documentation for FinanceModels.jl or the docstrings of FinanceModels.jl's avaiable model types.
+See the tutorials in the package documentation for FinanceModels.jl or the docstrings of FinanceModels.jl's available model types.
 """
 function fit(mod0, quotes, method::F=Fit.Loss(x -> x^2);
     variables=__default_optic(mod0),
@@ -281,6 +282,21 @@ function fit(mod0, quotes, method::F=Fit.Loss(x -> x^2);
 
 end
 
+function fit(mod0::Yield.MonotoneConvexUnInit, quotes, method;
+    variables=__default_optic(mod0),
+    optimizer=__default_optim(mod0)
+)
+    # use maturities as the times
+    # fit a vector of rates to the times
+    times = [maturity(q.contract) for q in quotes]
+    if !iszero(first(times))
+        pushfirst!(times, zero(eltype(times)))
+    end
+    mc = mod0(times)
+    rates = fit(mc, quotes, method)
+end
+
+
 function fit(mod0::Spline.BSpline, quotes, method::Fit.Bootstrap)
     discount_vector = [0.0]
     times = [maturity(quotes[1])]

src/model/Yield.jl

@@ -68,6 +68,7 @@ end
 
 include("Yield/SmithWilson.jl")
 include("Yield/NelsonSiegelSvensson.jl")
+include("Yield/MonotoneConvex.jl")
 
 ## Generic and Fallbacks
 """

src/model/Yield/MonotoneConvex.jl

@@ -1,3 +1,30 @@
+"""
+An unfit Monotone Convex Yield Curve Model will simply have 
+
+
+"""
+struct MonotoneConvex{T,U} <: AbstractYieldModel
+    f::Vector{T}
+    fᵈ::Vector{T}
+    rates::Vector{T}
+    times::Vector{U}
+    # inner constructor ensures f consistency with rates at construction
+    function MonotoneConvex(rates::Vector{T}, times::Vector{U}) where {T,U}
+        f, fᵈ = __monotone_convex_fs(rates, times)
+        new{T,U}(f, fᵈ, rates, times)
+    end
+end
+
+
+struct MonotoneConvexUnInit
+end
+
+MonotoneConvex() = MonotoneConvexUnInit()
+function (m::MonotoneConvexUnInit)(times)
+    rates = zeros(length(times))
+    MonotoneConvex(rates, times)
+end
+
 
 function __issector1(g0, g1)
     a = (g0 > 0) && (g1 >= -2 * g0) && (-0.5 * g0 >= g1)
@@ -13,13 +40,11 @@ end
 
 # Hagan West - WILMOTT magazine pgs 75-81
 function g(x, f⁻, f, fᵈ)
-    @show x, f⁻, f, fᵈ
-    @show g0 = f⁻ - fᵈ
-    @show g1 = f - fᵈ
+    g0 = f⁻ - fᵈ
+    g1 = f - fᵈ
     A = -2 * g0
     B = -2 * g1
     if sign(g0) == sign(g1)
-        @show "(iv)"
         # sector (iv)
         η = g1 / (g1 + g0)
         α = -g0 * g1 / (g1 + g0)
@@ -32,11 +57,9 @@ function g(x, f⁻, f, fᵈ)
 
 
     elseif __issector1(g0, g1)
-        @show "(i)"
         # sector (i)
         g0 * (1 - 4 * x + 3 * x^2) + g1 * (-2 * x + 3 * x^2)
     elseif __issector2(g0, g1)
-        @show "(ii)"
         # sector (ii)
         η = (g1 + 2 * g0) / (g1 - g0)
         if x < η
@@ -45,7 +68,6 @@ function g(x, f⁻, f, fᵈ)
             return g0 + (g1 - g0) * ((x - η) / (1 - η))^2
         end
     else
-        @show "(iii)"
         # sector (iii)
         η = 3 * g1 / (g1 - g0)
         if x > η
@@ -58,13 +80,11 @@ function g(x, f⁻, f, fᵈ)
 end
 
 function g_rate(x, f⁻, f, fᵈ)
-    @show x, f⁻, f, fᵈ
-    @show g0 = f⁻ - fᵈ
-    @show g1 = f - fᵈ
+    g0 = f⁻ - fᵈ
+    g1 = f - fᵈ
     A = -2 * g0
     B = -2 * g1
     if sign(g0) == sign(g1)
-        @show "(iv)"
         # sector (iv)
         η = g1 / (g1 + g0)
         α = -g0 * g1 / (g1 + g0)
@@ -77,11 +97,9 @@ function g_rate(x, f⁻, f, fᵈ)
 
 
     elseif __issector1(g0, g1)
-        @show "(i)"
         # sector (i)
         g0 * (x - 2 * x^2 + x^3) + g1 * (-x^2 + x^3)
     elseif __issector2(g0, g1)
-        @show "(ii)"
         # sector (ii)
         η = (g1 + 2 * g0) / (g1 - g0)
         if x < η
@@ -90,7 +108,6 @@ function g_rate(x, f⁻, f, fᵈ)
             return g0 * x + ((g1 - g0) * (x - η)^3 / (1 - η)^2) / 3
         end
     else
-        @show "(iii)"
         # sector (iii)
         η = 3 * g1 / (g1 - g0)
         if x > η
@@ -107,14 +124,8 @@ function forward(t, rates, times)
     t, i_time, rates, times = __monotone_convex_init(t, rates, times)
     f, fᵈ = __monotone_convex_fs(rates, times)
 
-    @show x = (t - times[i_time]) / (times[i_time+1] - times[i_time])
-    @show fᵈ, f
-    @show fᵈ[i_time+1], g(x, f[i_time], f[i_time+1], fᵈ[i_time+1])
+    x = (t - times[i_time]) / (times[i_time+1] - times[i_time])
     return fᵈ[i_time+1] + g(x, f[i_time], f[i_time+1], fᵈ[i_time+1])
-
-
-
-
 end
 
 """
@@ -123,22 +134,22 @@ end
 function __monotone_convex_init(t, rates, times)
     # the array indexing in the paper and psuedo-VBA is messy
     t = min(t, last(times))
-    times = collect(times)
-    rates = collect(rates)
+    # times = collect(times)
+    # rates = collect(rates)
     i_time = findfirst(x -> x > t, times)
     if i_time == nothing
         i_time = lastindex(times)
     end
-    if !iszero(first(times))
-        pushfirst!(times, zero(eltype(times)))
-        pushfirst!(rates, first(rates))
+    # if !iszero(first(times))
+    #     pushfirst!(times, zero(eltype(times)))
+    #     pushfirst!(rates, first(rates))
 
-    end
+    # end
 
     return t, i_time, rates, times
 end
 """
-    returns a pair of vectors (f and fᵈ) used in Monotone Convext Yield Curve fitting
+    returns a pair of vectors (f and fᵈ) used in Monotone Convex Yield Curve fitting
 """
 function __monotone_convex_fs(rates, times)
     # step 1
@@ -176,31 +187,54 @@ function myzero(t, rates, times)
     t, i_time, rates, times = __monotone_convex_init(t, rates, times)
     f, fᵈ = __monotone_convex_fs(rates, times)
     x = (t - times[i_time]) / (times[i_time+1] - times[i_time])
-    @show G = g_rate(x, f[i_time], f[i_time+1], fᵈ[i_time+1])
+    G = g_rate(x, f[i_time], f[i_time+1], fᵈ[i_time+1])
     return 1 / t * (times[i_time] * rates[i_time] + (t - times[i_time]) * fᵈ[i_time+1] + (times[i_time+1] - times[i_time]) * G)
 
 
 
 
+end
+
+function Base.zero(mc::MonotoneConvex, t)
+    lt = last(times)
+    # if the time is greater than the last input time then extrapolate using the forwards
+    if t > lt
+        r = myzero(lt, rates, times)
+        return r * lt / t + forward(lt, rates, times) * (1 - lt / t)
+    end
+
+    t, i_time, rates, times = __monotone_convex_init(t, rates, times)
+    f, fᵈ = mc.f, mc.fᵈ
+    x = (t - times[i_time]) / (times[i_time+1] - times[i_time])
+    G = g_rate(x, f[i_time], f[i_time+1], fᵈ[i_time+1])
+    return 1 / t * (times[i_time] * rates[i_time] + (t - times[i_time]) * fᵈ[i_time+1] + (times[i_time+1] - times[i_time]) * G)
+
+end
+
+function FinanceCore.discount(mc::MonotoneConvex, t)
+    r = zero(mc, t)
+    return exp(-r * t)
 end
 
 times = 1:5
 rates = [0.03, 0.04, 0.047, 0.06, 0.06]
-forward(5.19, rates, times)
-myzero(1, rates, times)
-
-
-using Test
-@test forward(0.5, rates, times) ≈ 0.02875
-@test forward(1, rates, times) ≈ 0.04
-@test forward(2, rates, times) ≈ 0.0555
-@test forward(2.5, rates, times) ≈ 0.0571254591368226
-@test forward(5, rates, times) ≈ 0.05025
-@test forward(5.2, rates, times) ≈ 0.05025
-
-@test myzero(0.5, rates, times) ≈ 0.02625
-@test myzero(1, rates, times) ≈ 0.03
-@test myzero(2, rates, times) ≈ 0.04
-@test myzero(2.5, rates, times) ≈ 0.0431375956535047
-@test myzero(5, rates, times) ≈ 0.06
-@test myzero(5.2, rates, times) ≈ 0.059625
+# forward(5.19, rates, times)
+# myzero(1, rates, times)
+
+
+# using Test
+# @test forward(0.5, rates, times) ≈ 0.02875
+# @test forward(1, rates, times) ≈ 0.04
+# @test forward(2, rates, times) ≈ 0.0555
+# @test forward(2.5, rates, times) ≈ 0.0571254591368226
+# @test forward(5, rates, times) ≈ 0.05025
+# @test forward(5.2, rates, times) ≈ 0.05025
+
+# @test myzero(0.5, rates, times) ≈ 0.02625
+# @test myzero(1, rates, times) ≈ 0.03
+# @test myzero(2, rates, times) ≈ 0.04
+# @test myzero(2.5, rates, times) ≈ 0.0431375956535047
+# @test myzero(5, rates, times) ≈ 0.06
+# @test myzero(5.2, rates, times) ≈ 0.059625
+
+